Neuraxial Analgesia In Neonates And Infants: Review of Clinical and Preclinical Strategies for the Development of Safety and Efficacy Data

Abstract

Neuraxial agents provide robust pain control, have the potential to improve outcomes, and are an important component of the perioperative care of children. Opioids or clonidine improve analgesia when added to perioperative epidural infusions; analgesia is significantly prolonged by addition of clonidine, ketamine, neostigmine or tramadol to single shot caudal injections of local anesthetic; and neonatal intrathecal anesthesia/analgesia is increasing in some centers. However, it is difficult to determine the relative risk-benefit of different techniques and drugs without detailed and sensitive data related to analgesia requirements, side-effects, and follow-up. Current data related to benefits and complications in neonates and infants are summarized, but variability in current neuraxial drug use reflects the relative lack of high quality evidence. Recent preclinical reports of adverse effects of general anesthetics on the developing brain have increased awareness of the potential benefit of neuraxial anesthesia/analgesia to avoid or reduce general anesthetic dose requirements. However, the developing spinal cord is also vulnerable to drug-related toxicity, and although there are well-established preclinical models and criteria for assessing spinal cord toxicity in adult animals, until recently there had been no systematic evaluation during early life. Therefore, the second half of this review presents preclinical data evaluating age-dependent changes in the pharmacodynamic response to different spinal analgesics, and recent studies evaluating spinal toxicity in specific developmental models. Finally, we advocate use of neuraxial agents with the widest demonstrable safety margin and suggest minimum standards for preclinical evaluation prior to adoption of new analgesics or preparations into routine clinical practice.

INTRODUCTION

The consequences of inadequate regulation of pain were made evident by early clinical studies showing that anesthesia and analgesia reduced morbidity and mortality following cardiac surgery in the newborn^1,2. As well as deleterious acute physiologic consequences, there is an evolving literature indicating that neonatal surgery and/or intensive care can result in prolonged changes in sensory processing^3–6 and altered responses to future pain^7–9. While adequate intra-operative anesthesia and analgesia in the newborn, as in the adult, can be achieved by inhalants and intravenous drugs, there has long been an appreciation of the benefits of neuraxial anesthetics and analgesics, which can create dense local anesthesia and analgesia that extends into the perioperative period with reduced systemic side effects. The use of neuraxial drugs in the control of pain may now be further encouraged as recent data demonstrate that general anesthetics (NMDA antagonists, isoflurane, nitrous oxide) and benzodiazepines produce developmentally regulated increases in perinatal apoptosis and long term deleterious behavioral changes ^10–12. However, it is important to appreciate that neuraxial delivery employs agents which until recently have never been systematically assessed for their safety during early development. This has been highlighted by the Anesthetic and Life Support Drugs Advisory Committee of the FDA^*, which stated that “the potential for anesthetic agent-induced neurodegeneration at the level of the spinal cord should be evaluated, particularly with respect to the local anesthetics and opioids administered neuraxially”.

An increasing number of drugs and preparations have been used to produce neuraxial analgesia, with clinical studies demonstrating tolerability and efficacy. However, high quality evidence for improved clinical outcomes, particularly in neonates and infants, is limited. There is a growing emphasis on the need for preclinical evaluation of spinal toxicity to fully evaluate the relative benefits and risks of different agents prior to clinical use. This is reflected by the adoption of specific guidelines for publication of neuraxial clinical trials by several major journals. In this review, we seek to address four specific issues: 1) summarize the clinical use of neuraxial techniques in neonates and infants; 2) highlight current difficulties in evaluating the comparative benefit and potential risk of different spinal analgesic drugs; 3) summarize preclinical models evaluating developmental changes in the pharmacodynamic response to spinal analgesic drugs; and 4) review minimal standards for implementation of spinal agents in neonates to permit informed assessment between different agents in terms of efficacy and toxicity in the neonate. The review will consider agents that block conduction (i.e. local anesthetics), but will focus on those that specifically attenuate the spinal processing of pain information when administered by the intrathecal or epidural/caudal route (i.e. spinal analgesics, also often termed spinal adjuvants).

1. CLINICAL USE OF NEURAXIAL ANALGESIA AND ANESTHESIA IN NEONATES AND INFANTS

Neuraxial delivery

The control of afferent traffic through neuraxial interventions (epidural or intrathecal delivery) can be utilized in neonates and infants as (i) a sole neuraxial anesthetic technique for abdominal and lower limb surgery ^13,14; or (ii) as a supplement to reduce intraoperative general anesthetic requirements and manage peri-operative pain^15,16.

Intrathecal delivery of local anesthetic produces “spinal” anesthesia. Use of neonatal spinal anesthesia is increasing in some centers^17,18, with large series reporting safe and effective anesthesia and analgesia^13,19,20, including use in high risk and extremely low birth weight neonates²¹. “Single shot” spinal anesthesia provides an alternative to general anesthesia for lower abdominal or inguinal surgery, however the clinical utility of this technique is limited by the duration of action of intrathecal local anesthetics in neonates, and conversion to general anesthesia is often required if surgical duration exceeds one hour^13,19. Various techniques have been utilized in infants and neonates to prolong the duration of intrathecal anesthesia including: (i) repeat administration via an intrathecal catheter²²; (ii) a combined spinal and epidural catheter (CSE) technique for upper abdominal surgery^23,24; (iii) additional local anesthetic administration by the surgeon during myelomeningocoele surgery; and (iv) addition of spinal analgesic adjuvants such as opioids^22,25 or clonidine^26,27.

Epidural analgesia can also be used as a sole technique ^28,29 or as a supplement to general anesthesia for perioperative analgesia ³⁰ for neonatal and infant surgery. Single bolus administration ¹⁵, or infusion via a catheter advanced from the caudal space ³¹ or inserted at an intervertebral level in the thoracic or lumbar spine ³² is possible in even the smallest preterm neonate ³³. A range of spinal analgesics are now administered, often in conjunction with local anesthetics, with the aim of: (i) improving analgesia; (ii) reducing local anesthetic requirements and associated side-effects such as motor block; and (iii) prolonging analgesia following single shot administration.

Epidemiology

Neuraxial analgesia is used in children of all ages, but the pattern of use and choice of technique varies with age of the child, across institutions, and with time in some centers^18,34. In a 1994 survey of regional anesthesia by the French-language Society of Pediatric Anesthesiologists (ADARPEF), neonates comprised 3.3% of the pediatric population, but received 3.4% of caudals, 1.8% of epidurals, and 10.9% of all spinal anesthetics.³⁴ A similar survey in 2006 found a decrease in the use of caudals, but increased use of epidural catheters and single shot spinal anesthetics, and a greater proportion of central blocks were being performed at younger ages (5.6 vs 3.4% in neonates; 30 vs 16.5% in infants < 6 months).¹⁷ In one French center, the overall proportion of neuraxial blocks decreased from 1989 to 2005, but spinal anesthetics in neonates had become the most frequent technique, comprising 30% of the total¹⁸. The number of epidurals performed annually in UK children was stable from 2002 to 2005, with 5% of the total 10633 epidural performed in neonates and 16% in children aged between 1 month and 1 year ³⁵.

2. CLINICAL BENEFITS AND RISKS OF NEURAXIAL ANALGESIA

Potential advantages of neuraxial route in neonates and infants

In addition to minimizing the potential exposure of the developing brain to general anesthetics, neuraxial analgesia may improve postoperative outcomes for high-risk neonates who are susceptible to respiratory complications (e.g. preterm born neonates with lung disease and postoperative apnea) ³⁶ or who require major surgery for correction of congenital anomalies ^37–40. However, the magnitude of benefit of intra-operative or peri-operative neuraxial anesthesia is difficult to determine from case reports or series ⁴¹. Even in older children undergoing scoliosis surgery, meta-analysis demonstrated improved analgesia with epidural local anesthetic and opioid versus systemic opioid for adolescents, but there was insufficient data to confirm any change in respiratory outcomes, length of hospital stay, or mortality ⁴². In younger children, variability in study design (type of surgery; neuraxial anesthesia regimes with local anesthetic in varying concentrations and doses and different types and doses of spinal analgesic) makes systematic analysis of outcomes even more difficult. Reported benefits of neuraxial anesthesia in studies that include neonates and infants are:

Reduction in respiratory complications

1. post-operative apnea. Analysis of four trials comparing spinal and general anesthesia in neonates born preterm undergoing inguinal herniorrhaphy found a reduction in the incidence of postoperative apnea only if systemic sedatives were avoided ³⁶. Neuraxial anesthesia and avoidance of opioids may have added advantages in neonates with central hypoventilation syndromes ⁴³. It has been suggested that spinal anesthesia can reduce costs related to postoperative monitoring and hospitalization ⁴⁴.

2. post-operative mechanical ventilation. In a randomized trial of infants undergoing cardiac surgery, caudal morphine and local anesthetic provided some analgesic benefits over systemic morphine, but the study had insufficient power to evaluate effects on early extubation ⁴⁵. In case series comparing perioperative neuraxial anesthesia with systemic opioid analgesia, the proportion of neonates requiring postoperative mechanical ventilation was reduced following gastrochisis repair ³⁸, lung resection for congenital lung lesions ⁴⁶, and Nissen fundoplication ⁴⁷. Cases of improved respiratory function following major neonatal thoracic surgery with epidural analgesia have been reported ^48,49.

Attenuation of stress response

Circulating levels of stress hormones such as cortisol ^50,51, adrenaline and noradrenaline ^22,52 are reduced when supplementary neuraxial anesthesia is added to general anesthesia.

Cardiac stability

Maintenance of cardiovascular stability has been demonstrated during neuraxial techniques in neonates ⁵³, including CSE anesthesia for upper abdominal surgery ²⁴ and in high-risk neonates and infants with congenital cardiac disease ⁵⁴. While these observations support the safety of the technique, improved outcomes in comparisons with general anesthesia have not been confirmed.

Reduction in hospital stay

In uncontrolled trials, epidural rather than systemic analgesia reduced hospital stay following ligation of patent ductus arteriosus in infants ⁵⁵ and fundoplication ⁴⁷.

Improved surgical outcome

Wound dehiscence following bladder exstrophy repair in neonates was avoided with prolonged neuraxial anesthesia (mean 15 days) and sedation, but there was no comparison with other analgesic techniques ³⁷.

Potential disadvantages of neuraxial route in neonates and infants

Complication Rates

Although severe complications following pediatric neuraxial techniques are rare, the incidence is higher in neonates and infants: 0.4% vs 0.1% for all neuraxial blocks ¹⁷ and 1.1% vs 0.49% for epidural blocks alone ³⁵. Outcomes may be worse in neonates ^56,57, with complication rates as high as 4:1000 (including 3 deaths) ⁵⁶ initially reported, but more recent surveys report complication rates of 0.29% (95%CI: 0.21–0.43) for central blocks (caudal, epidural and spinal; n=10,556) ¹⁷. Following peri-operative epidural infusions (n=10,633), the rate of serious incidents approximated 0.5:1000, with an additional 0.75:1000 incidents graded as moderate severity ³⁵.

The clinical practice setting, resource availability, and experience of individual practitioners can have a major impact on the relative risk and benefit of neuraxial anesthesia. The lack of intensive care facilities in some practice settings will increase the potential benefit of neuraxial techniques that reduce the requirement for postoperative mechanical ventilation. Management by experienced practitioners may minimize the incidence and severity of adverse events in neonates, as skilled intraoperative resuscitation was required following dural puncture or intravascular injection ³⁴. Complications related to the use of wrong equipment (eg. inappropriate or oversized needles, excessive length of catheter introduced into space) were reported in early series ³⁴. Pump programming or prescription errors were more common in young children (0.3% in children under 1 year versus 0.07% in 1–8 year-olds) ³⁵. All were corrected before harm occurred, but this emphasizes the need for adequate monitoring and follow-up of patients with epidural infusions

Infection

Asymptomatic colonization of epidural catheters is common (35%) but in series of 210 ⁵⁸ or 1458 children ⁵⁹, no local or systemic infections were reported. Age was not a clear factor, although the rate of colonization was higher for caudal than lumbar catheters in the under 3 year age group ⁵⁸. In a national audit of 10,633 perioperative epidural infusions, there were 25 cases of local skin infection (ages not reported); epidural abscess was reported in 2 cases (including one infant), and an additional 16 year old patient developed signs of meningism ³⁵. In a single center over 17 years, epidural catheter related infection, limited to the paraspinal or subcutaneous tissue, occurred in 6 of 10,437 (0.06%) cases, including one neonate and one infant ⁶⁰. All presented with back pain, pyrexia, and cellulitis; 5 also had pus visible at the catheter exit site; 3 required surgical drainage; and all recovered without neurological sequelae. Epidural catheters inserted for longer periods for chronic pain management were associated with higher rates of infection (3.2% vs 0.06%) ⁶⁰.

Neurological injury

Rates of neurological injury following neuraxial analgesia range from 0.13 to 0.4 per 1000 in large series, with higher rates following epidural catheter techniques than single shot caudals. Transient neuropathy was reported following 2 per 15,013 central blocks³⁴ and 6 per 10,633 epidural anesthetic infusions³⁵. In addition, following a programming error that rapidly delivered 15ml of solution, a 4-month-old preterm born infant developed cauda equina syndrome with persisting neurological deficit one year later³⁵. Suspected nerve injuries occurred following 1 of 364 thoracic, 2 of 1183 lumbar, and 1 of 8493 caudal epidural blocks, with no reported long-term deficits, and children were aged 8 years and above¹⁷. Isolated cases of neurological deficit following neuraxial anesthesia of varying severity have been reported in neonates⁶¹ and older children^62–65. The relative contributions of needle trauma, surgical injury, or potential drug-related toxicity to neurological injury are difficult to determine. No neurologic sequelae were reported in a retrospective review of 750 children (52% of whom were infants) requiring cardiac surgery and treated with peri-operative epidural local anesthetic, opioid and/or clonidine ⁶⁶. However, as in many studies, the duration of follow-up and the nature and sensitivity of neurological evaluation was not reported. The rates of complications may be under-estimated, particularly in young children ⁶⁷, who cannot report sensory symptoms and subtle motor changes are difficult to detect in infants not yet walking. More thorough follow-up of patients following neuraxial blocks has been advocated ⁶⁸.

CLINICAL CHOICE OF A SPINAL ANALGESIC: EFFICACY

Local anesthetics

The primary drugs delivered neuraxially in neonates are local anesthetics and examples of the range of preparations used in neonates and infants are included in Table 1. Issues of safety with neuraxially administered local anesthetics have tended to focus on systemic toxicity and high plasma concentrations that precipitate neurological and cardiovascular complications (i.e. convulsions and arrhythmias) ^69,70. Age-related alterations in pharmacokinetics result in higher free drug concentration following a bolus and accumulation of local anesthetic during infusion in neonates ^71–75. As a result, infusion duration tends to be limited in the youngest patients. In a recent study of neonates following bladder exstrophy repair, epidural lidocaine was infused for an average of 15 days (range 4–30 days), but with regular monitoring of plasma lidocaine concentration ³⁷. As will be reviewed below, it should be emphasized that although widely employed, there have until recently been no systematic studies as to potential adverse effects upon the developing spinal cord ⁷⁶, and no comparative studies of different local anesthetics.

Table 1.

Spinally-administered Local Anesthetics in Neonates and Infants.

	Route	Concentration/Dose	Age	Design/sample	Outcome/Results	Side-effects/Complications	Ref
Bupivacaine	IT	0.5mg/kg 0.5% = 0.1ml/kg (<5kg); 0.4mg/kg (>5kg)	< 1mth n=20; 1–3mths n=26; 3–6mths n=22	Case series; inguinal hernia repair	In youngest: higher proportion (95%) with adequate spinal but less postop analgesia	Mild hypotension	288
	IT	Hyperbaric 0.5% 1mg/kg + epinephrine	Birth 24–40wks; postnatal age 5–24wks, n=20	Case series: 17 inguinal hernia, 2 stoma, 1 teratoma; BIS monitoring	100% successful block; decrease BIS values 15 mins post spinal	>20% decrease in BP; HR stable	289
	IT	Isobaric 0.5% bupi vs hyperbaric 0.5%bupi in 8% glucose; 0.5mg/kg <10kg, 0.4mg/kg 11–19kg, 0.3mg/kg >20kg	2–115 mths, n=100	DB, randomized; lower abdo and lwr limb	Success (complete sensory block) higher with hyperbaric (95% vs 82%); no difference in height of sensory block (but wide variability both groups), degree motor block or postop analgesia	Treatment required: 10 suppt O2; 1 hypotension; 1 bradycardia	290
	IT	Isobaric 0.5%; mean dose 0.68±0.16mg/kg	Gestational age at operation: 47.6 (28–120) wks; n=505	Case series: lwr abdo, perineal, lwr limb (79% ing hernia)	Successful spinal 96%; conversion to GA 1% (surgery >90mins); sedation required 28%	Ave. spinal tap attempts 1.4 (1–6); bloody tap 12.4%; Bradycardia 1.8% (6/9 require atropine); High block 3pts (2 req. intubation)	19
	IT + EP-B (C)	Isobaric 0.5% 1mg/kg IT in spinal grp; All: plus caudal 0.25% 2mg/kg	40 (36–44) wks PCA; n=10 (n=14 GA sevoflurane)	RCT: spinal vs GA	Successful spinal 72%; decreased postop cardiorespiratory events in spinal grp	Unsuccessful spinal 28% (4/14)	291
	IT (CSE)	IT: isobaric 1mg/kg 0.5% Caudal catheter (advanced to thoracic): 0.2mg/kg/hr 0.1%	31–53 wks PCA, n=28	Case series; CSE sole technique for major upr abdo surgery	Satisfactory surgical anesthesia in 24/28 (4 convert to GA); 20 supplemental midazolam;	Multiple spinal attempts 3/24; CVS parameters stable	24
	EP-B (L)	Iml/kg 0.25% bupi+epinephrine then 1 ml/kg 0.125% every 2 hrs intra-operatively	36–41 wks PCA repair on 4th±5 days (1–23dys) n=23	Case series: bladder exstrophy repair	Intraoperative bupivacaine bolus (postop lidocaine; see below). 7/23 required intraop fentanyl.		37
Ropivacaine	IT	1.08mg/kg 0.5% = 0.216ml/kg (ED95)	<55 wks PMA n=50	Dose finding, inguinal hernia	Effective; motor block shorter duration and variable (cf other agents)		292
	EP-B EP-I	0.9–2mg/kg 0.2% (L) + 0.2mg/kg/hr 0.2% (<6mths); 0.4mg/kg/hr (>6mths)	0–1mth, n=11; 1–3mths, n=10; 3–6mths, n=10; 6–12mths, n=14	Case series; major abdo/thoracic surgery	Decreased clearance in neonates; Unbound plasma concentration higher in neonates		293
	EP-I	0.2% 0–2ml/kg/hr	Neonate; n=22	Case series; gastroschisis	Decreased postop ventilation in regional vs opioid		38
	EP-B (C)	1ml/kg 0.2%	0–12mths, n=10; 1–5yrs, n=10	Pharmacokinetic study; inguinal hernia repair	Higher plasma concentration in infants	No signs systemic toxicity	294
Levo-bupivacaine	IT	0.5 to 1.2 mg/kg 0.5%	< 55 wks PCA; n=50	Dose finding, lower abd surgery	Recommended dose 1mg/kg	No significant adverse effects	295
	EP-B (C)	2 mg/kg 0.25%	2±0.7 (0.6–2.9) months; n=22	Pharmacokinetic study; lwr abdo surgery (82% inguinal hernia)	Decreased clearance in infants		296
Lidocaine	EP-I	0.8–1mg/kg/hr 0.1% postoperative infusion	36–41 wks PCA repair on 4th±5 days (1–23dys) n=23	Case series: bladder exstrophy repair *Duration: 15±8 days (4–30days)	Adjust infusion to maintain plasma concentration <5mg/L; 22/23 required reduction in infusion in first 48hrs. Tunneled catheter: 10/23 early dislodgement at 13±7 (6–28) days		37
Tetracaine	IT	Lidocaine 3mg/kg + epinephrine Tetracaine 0.4mg/kg or 0.4mg/kg + epinephrine	1mth–12 mths (7/100 < 44 wks PCA; 77/100 < 6mths); n=100	Case series: lwr abd, lwr limb (87% inguinal hernia)	Duration motor block: lido + epin 56±2.5 mins tetracaine 86±4mins tetracaine + epin 128±3mins. Sedation: methohexital 78; ketamine 4)	Four or less spinal tap attempts; no bloody taps. CVS stable. Follow-up 7–10days: no complications	297
	IT/CSE	Mean dose 0.65mg/kg Post-op caudal catheter: bupivacaine 0.25mg/kg/hr (neonates) or 0.5mg/kg/hr (infants)	29wks PCA to 7 months; 1.5–7.8kg; n=19	Case series; major abdo surgery	+ CSE bupi via caudal cath	Sedation 7/19 (1 propofol, 6 midazolam); Subarachnoid block with catheter 1 (required intubation)	23
	IT	Mean dose 0.56 mg/kg 0.5% in dextrose 5%	Neonates, n=14	Case series: repair meningomyelocele	Additional doses by surgeon	Postop apnea in 2 with midazolam sedation	298
	IT	1ml/kg 0.5% tetracaine (or 0.5% bupi) + 5% glucose + adrenaline	24–42 wks gest age, n=62	Case series; 58 ing hernia, 3 pyloromyotomy, 1 muscle Bx	Spinal success 89%;	5.4% req GA; apnea 3%; bradycardia 4%	299
	IT	0.5ml/kg 0.5% tetracaine + 5% dextrose + epinephrine	24–37 wks gest age, n=142	Case series: 95% inguinal hernia; 5% urology	Spinal success 96%;	4.5% sedation; apnea 0.8%; brady 1.5%	44
	IT	Hyperbaric; mean dose 0.54±0.2mg/kg (+epinephrine in 91% (excluding PDA); [0.4% cases: hyperbaric bupi or lidocaine]	Neonates and infants (<12 mths): 650g–13kg; n=1554	Case series: abdo, lwr abdo and lwr limb; urology; myelomeningocoele; (55% ing hernia)	Spinal adequate for surgery 95%; supplemental LA by surgeon 2.7%;	1.4% conversion to GA (surgery duration > block); bradycardia 1.6% (15/24 require treatment); sedation 24%; SaO2 <90% 0.6%; high block 56 patients (5 assist ventilation, 5 intubated)	13
	IT	Hyperbaric 0.5% in 5% dextrose + epinephrine; Mean dose 2.4mg/kg	Neonates mean PCA 33 (28–41) weeks; 1276 (650–2965)g n=14	Case series, PDA repair	Intubated and high dose to aim for total spinal; CVS stable;	Supplement 7/14 (isoflurane, midazolam, N2O or fentanyl); Decrease BP & HR 1/14	40
Chloro-procaine	EP-B+I (C)	1ml/kg 3% ± 0.3ml/kg bolus to establish + 1ml/kg/hr	Ex-preterm 35–49 wks PCA; n=10	Case series; feasibility in awake, inguinal hernia (sole)	Mean cumulative requirement 2.8±1 ml/kg/hr	BP mild decrease; one apnea (pre-existing episodes)	28
	EP-B+I (C)	1–1.5ml/kg 3% bolus + 1–1.5 ml/kg/hr	Neonates, 1–28 days; 2.2–4.9kg; n=25	Case series, major abdo surgery (GA suppt)	CVS stable	Caudal space on 1st or 2nd attempt; CVS stable	29

Legend: IT: intrathecal; EP-B: epidural bolus administration; EP-I: epidural infusion; (L): lumbar injection; (L/T): lumbar or thoracic insertion/injection; (C): caudal injection; n.s. not statistically significant

NB: studies with LA combined with opioid or adjuvant reported in Table 2

Spinal analgesics and clinical study design

Few studies have directly compared the efficacy of different spinal analgesic drugs in children of different ages. This, and the lack of systematic safety data (discussed below), makes it difficult for practitioners to make an evidence-based choice between different drugs, thus contributing to the wide variability in current clinical practice ^77,78.

Evaluating data from current controlled trials is hampered by variation in methodology, particularly in the sensitivity of the outcome measures and end-points used to measure the duration and efficacy of analgesia. In neonates and infants, sample sizes are frequently small ^79,80 as recruitment of large homogeneous samples is difficult, and may be further constrained by ethical issues ⁸¹. Additional variability in the type, sensitivity and specificity of pain assessment tools utilized ⁷⁸ may further reduce the power of the study.

Prolongation of analgesia

If analgesia is being titrated against individual requirements, differences in pain scores should not be seen, and therefore differences in the duration of analgesia or supplemental analgesia requirements are often used as outcome measures. The most frequent comparison is between the same dose of local anesthetic with or without a spinal analgesic, and relatively few studies evaluate the ability of spinal analgesics to reduce the required concentration of local anesthetic ^82,83 or the impact of different doses of local anesthetic ⁸⁴. Time to first analgesia will be influenced by: the sensitivity, frequency and inter-rater reliability of pain assessment (particularly following discharge when reliance is placed on parental interventions); the trigger for administration; and the type of supplemental analgesic. Meta-analyses have demonstrated statistically significant prolongation of analgesia with caudal clonidine ^79,85,86 and ketamine ^84,87. The remaining question is whether the degree of change is clinically, as well as statistically, significant. As reported increases in duration range from 2.3 to 5.3 hours, analgesia may be receding soon after the patient leaves the PACU or when ambulatory patients are leaving the hospital, and this needs to be considered when providing instructions to ward staff and parents regarding supplemental analgesia.

Supplemental analgesia

The clinical significance of a reduction in supplemental analgesia as an outcome depends on the total dose, side-effect profile and relative risk of the different treatments. A reduction in opioid requirement with addition of spinal analgesics⁸⁸ has the potential to reduce opioid-related side-effects such as nausea and vomiting. However, many pediatric studies have been conducted following day case surgery, where postoperative pain scores and/or analgesic requirements are low, making it difficult to demonstrate a difference between two active treatments⁸⁹. A reduction in the use of mild analgesics such as acetaminophen or NSAIDs^84,87 provides evidence of an analgesic effect, but the relative risk of the spinal adjuvant must be weighed against that of the additional supplemental analgesia. We would question whether avoiding one or two doses of acetaminophen over a 24-hour period justifies the risk of neuraxial administration of a drug that has not been evaluated for spinal toxicity. In addition, studies may report only the proportion of children requiring analgesia, or the total number of doses in the whole treatment group, and therefore dose requirements and relative benefits or risks for individual patients cannot be assessed.

Route of administration

Neuraxial analgesic administration has the potential to produce analgesia at doses lower than required with systemic administration, thus reducing side-effects. Epidural morphine (12–50mcg/kg) improves analgesia ^90–92, and although early systemic absorption was detected, analgesia was evident 1 and 3 hours later when plasma levels were lower than required for a systemic analgesic effect ⁸⁸. Lower doses (2–5μg/kg) are effective intrathecally ^93–95. The degree of dose sparing depends on the chemical properties of the drug, and for more lipophilic opioids such as fentanyl, the difference between equi-effective intrathecal, epidural and systemic doses may be less⁹⁶. Minimal dose sparing has also been demonstrated with ketamine, as 0.5 to 1mg/kg is utilized in caudal studies ^97,98 and the same dose systemically provides procedural sedation and analgesia ^99–101, albeit for a shorter duration ¹⁰². Similarly, analgesia was prolonged when comparing caudal and intravenous administration of 2mg/kg tramadol ¹⁰³. Clonidine via the intrathecal ¹⁰⁴ or caudal ¹⁰⁵ route has a greater effect on analgesic duration than the same dose intravenously, but effects on general anesthetic requirements and early post-operative sedation are seen with neuraxial and systemic administration.

Addition of caudal adjuvants following unilateral hernia repair in children often aims to reduce local anesthetic requirements and associated motor block, but less invasive techniques such as local infiltration and ilioinguinal block are also effective in the early postoperative period ⁷⁸. Few studies have directly compared different local anesthetic techniques. When compared to dorsal penile block for circumcision, caudal bupivacaine plus ketamine was found to have no advantage ¹⁰⁶, or to produce mild prolongation of analgesia (7.6 vs 6.2 hrs) at the cost of increased motor block ¹⁰⁷.

Spinal analgesic drugs

In the following section we will provide a commentary on the use of analgesics that are delivered by the intrathecal or epidural/caudal route, with the aim of producing spinally-mediated analgesia (i.e. spinal analgesics or spinal adjuvant analgesic drugs), and which are typically used in conjunction with local anesthetics. Table 1 provides a systematic summary of the reported literature relevant to the several families of adjuvant analgesics. In each case, the reported dosing is provided. In many cases there is limited information related to the concentration of the different drugs within the injectate, but when co-administered with local anesthetic, the desired spread and volume of local anesthetic is often the deciding factor.

Opioids are the most frequently utilized spinal analgesics, but increased knowledge of spinal pharmacology has led to drugs such as alpha-2 adrenergic agonists (clonidine), NMDA antagonists (ketamine), GABA agonists (midazolam) and neostigmine being used alone or in combination as spinal analgesics in adults ¹⁰⁸. Use of spinal analgesics has expanded to pediatric practice, but there is marked variability in the availability of different preparations and in the clinical use of these drugs. Surveys of pediatric anesthetists in the UK reported that 16% added clonidine, 15% ketamine and 9% epinephrine to epidural infusions ⁷⁷. The proportion using clonidine as a caudal analgesic has increased (26% in 2002 and 42% in 2009), whereas use of ketamine and midazolam remained relatively constant at 32–37% and 0.5–1% respectively^109,110. A survey of 25 international pediatric centers found an increase use of clonidine (18 to 23 of 25 centers) whereas use of ketamine had significantly decreased from 12 to 4 centers ¹¹¹. While the majority of controlled trials have been conducted in children over 6 months of age ⁷⁹, many spinal analgesics have been used in neonates and infants less than 6 months (see Table 2), despite limited evaluation of age-related changes in the pharmacodynamic profile of these drugs and no systematic evaluation of toxicity in the developing spinal cord.

Table 2.

Spinally-administered Analgesics in Neonates and Infants (≤6 months)

	Route	Concentration/Dose	Age	Design/sample	Outcome/Results	Complications	Ref
Opioids
Morphine	IT catheter	20mcg/kg bolus + 3mcg/kg/hr postop (with 0.1ml/hr 0.125% bupiv)	2–11 mths; cardiac surgery with CPB; n=30 (spinal grp)	RCT: spinal vs systemic opioid	Decreased stress response in spinal group; no difference in time to extubation	CVS stable; no spinal complications	22
	IT	7mcg/kg (with 2mg/kg tetracaine)	3mths – 6yrs, n=20 spinal group	RCT: spinal vs systemic opioid	Lower pain scores and decreased postop fentanyl requirement (no diff in opioid side-effects)	CVS stable	25
	EP-B (C)	100mcg/kg (with bupi 0.25% 1ml/kg + epinephrine 1:200000)	3 – 56 mths; n=63 (31/32 per grp)	RCT: pre-incision caudal vs postop IV opioid; cardiac surgery (single ventricle palliation)	No difference in extubation rates; postop morphine reqt n.s.;	No adverse events specifically related to caudal	45
Hydro-morphone	EP-B+I	5–10mcg/kg + 0.6–1.5 mcg/kg/hr	n=220 (57 infants <12 months)	Case series cardiac surgery (range of drugs and techniques; not clear if all used in infants)	Regional safe and effective in cardiac surgery	Intravascular puncture 1; paraesthesia 7 (?age); no identifiable spinal hematoma on postop neurological examination	300
Dia-morphine	EP-B (C)	30mcg/kg plus bupi 0.25% 0.5ml/kg VS bupi alone	6 – 88 months; n=45	RCT; hypospadius repair	Reduced pain scores first 30mins postop	Minor decrease RR at 15mins; PONV n.s. difference (3/22 vs 1/23)	120
Fentanyl	EP-I	Bolus 2mcg/kg + 0.21mcg/kg/hr (+0.2mcg/kg/hr bupivacaine)	12 days – 18 years n= 348 (87 <2yrs)	Case series; 80% orthopaedic	Effective analgesia	Cardiorespiratory stable; back pain at puncture site 1 (?age); fever and catheter removal 11 (tip culture all negative); mechanical problems and early cessation 25; urinary retention (without routine catheter) 17%	301
	EP-B+I (L)	1–2 mcg/kg bolus + 0.2mcg/kg/hr (plus bupivacaine 0.25% with epinephrine (up to 0.8ml/kg bolus) + 0.1% 0.2ml/kg/hr for mean 45 (39–69) hrs	Neonates; n=14	Case series, major abdo surgery; mean duration 43.7±8 hrs	Satisfactory analgesia in all	Dural puncture 1/14	30
	IT	0.25, 0.5 or 1mcg/kg (plus 0.5% hyperbaric bupivacaine)	Infants, mean 6–7 months; n=42	RCT; lwr abdo and urology	Addition 1mcg/kg fentanyl prolonged duration of SA block (74±6 vs 51±5 mins) and reduced postop rescue analgesia; 0.25 and 0.5mcg/kg not different from bupi alone	All low dose propofol infusion, ceased in 4 as required assisted ventilation; pruritus 3/42	302
Clonidine	IT	1mcg/kg clonidine plus hyperbaric bupi 0.5mg/kg VS 1mcg/kg fentanyl plus bupi VS 1mcg/kg clon plus fent plus bupi VS bupi alone	Infants 2–11 months (ex-preterm excluded); n=61 (15–16 per grp)	RCT DB; lwr abdo surgery under SA block (80% inguinal hernia)	Sensory block height T4–T8	Sedation score higher and intraop propofol requirement lower in BC & BCF grps; CVS stable	303
	IT	0.25, 0.5, 1 or 2mcg/kg (plus 0.5% isobaric bupi 1mg/kg)	38–46 wks PCA, n=75	RCT; inguinal hernia	Duration incr by 1 & 2mcg/kg; recommend 1mcg/kg	MAP decr by 22–40% (higher proportion MAP<40mmHg in C2); HR decr 12–27% all groups; no diff early apnea **limited follow-up until PACU discharge	26
	IT	1mcg/kg (plus 0.5% isobaric bupi 1mg/kg)	Prem vs term (29–50wks vs 39–53wks current PCA), n=67+57 = 124	Prospective observational; inguinal hernia	Success rate 84%	Unsuccessful block 10; inadequate duration 13; high block and resp impairment 1. Incr proportion apnea postop (6 before surg, 26 in 24hrs post surgery); increased bradycardia postop 7; “deeply sedated” 5	27
	EP-I (L/T)	Bolus 2mcg/kg + 0.2mcg/kg/hr OR 0.2mcg/kg/hr plus ropivacaine 0.1% (0.2mls/kg/hr)	3 – 98 months; n=35	Randomised, non blinded; major abdo surgery	“good analgesia” in both groups;; rescue analgesia required for cough and movement	Clonidine bolus increased sedation and hypotension; HR and RR stable	132
Ketamine	EP-B (C)	s-ketamine 1mg/kg ± 1 or 2mcg/kg clonidine (3 groups)	1 – 72 mths (mean 26±24mths)	RCT; inguinal hernia	K+C longer duration; suppt analgesia in 24hrs: 63% vs 16% with combination (paracetamol ? single dose)	CVS stable; “no adverse CNS effects” (?criteria); 24 hr follow-up	98
	EP-B (C)	s-ket 0.5mg/kg ± 1ml/kg levobupi 0.15% or 0.175% or 0.2%	3 mths – 6 yrs (mean 3 yrs), n=164 (52–56 per grp)	RCT D-B; lwr abdo or urology (57% inguinal hernia)	Adequate analgesia on incision 162/164 0.175% + K: lwr analgesic reqt (22/52 vs 38/56 vs 30/56)	Postop agitation 34/164; no ‘excess agitation or odd behavior’; 6hrs postop follow-up	97
	EP-B (C)	Bupi 0.125% 0.2ml/kg ± s-ketamine 0.5mg/kg	1 mth – 9yrs (mean 2.7yrs), n=30	RCT D-B; lwr abdo or urology (60% inguinal hernia)	10/15 in ket grp vs 3/15 no additional analgesia	CVS stable, no emergence delerium or unexplained distress	137
	EP-B (C)	s-ketamine 0.5mg/kg vs 1mg/kg vs bupi 0.25% 0.75ml/kg with epinephrine	3 mths – 6 yrs; n=49	RCT D-B; inguinal hernia repair	ket 1mg/kg = LA > 0.5mg/kg; 33 vs 30 vs 72% suppt paracetamol	CVS stable; no difference in sedation; no motor block at 6hrs	144
	EP-B (C)	0.25 or 0.5 or 1mg/kg plus bupi 0.25% 0.75mls/kg	6 mths – 10 yrs; n=60	RCT; unilateral inguinal hernia	0.5 and 1mg/kg prolonged analgesia and reduced rescue analgesia	1mg/kg increased behavioral side-effects 9/20 (odd, agitation, restless)	139
Neo-stigmine	IT	0, 0.25, 0.5, 0.75 or 1mcg/kg neostigmine plus hyperbaric bupi 0.5%	1 – 12 months; n=73 (14–15 per group)	RCT D-B; lower abdo or urology (55% inguinal hernia)	0.75 and 1mcg/kg reduced pain score and prolonged block duration;	CVS stable; emesis (10/73) did not differ across groups; apnea 6/73 (assist ventilation and cease propofol)	171
	EP-B (C)	2mcg/kg or 4mcg/kg plus levobupi 0.25% 1ml/kg vs levobupi 0.25% 1ml/kg	5 mths – 5 years; n=60	RCT: lwr abdo or inguinal (45% inguinal hernia)	2 and 4mcg/kg decrease pain score to 24hrs, prolong analgesia	PONV 4mcg/kg 3/20 (n.s.); CVS stable	162

IT: intrathecal; EP-B: epidural bolus administration; EP-I: epidural infusion; (L): lumbar injection; (L/T): lumbar or thoracic insertion/injection; (C): caudal injection; n.s. not statistically significant

Opioids

Mu opioids have been administered by epidural bolus and/or infusion and also as an intrathecal additive with local anesthetic. Morphine or fentanyl has been used most frequently in neonates and infants ^22,25,30, but the use of a wide range of opioid drugs has been reported in children 6 months and older including: alfentanil ¹¹², sufentanil ^113–115, buprenorphine ¹¹⁶, butorphanol ^117–119, diamorphine, ^120,121, hydromorphone ¹²² and tramadol ^{103,123–127}. In surveys of UK pediatric anesthetists, 85% used opioids for epidural analgesia⁷⁷, but variability in the agent chosen (fentanyl, morphine, or diamorphine) was noted in this and an earlier survey (21% adding fentanyl and 13% adding diamorphine to caudal anesthetic blocks) ¹⁰⁹. Although many practitioners had a minimum age for the use of epidural opioids, the cutoff varied from the neonatal period to 5 years of age ⁷⁷.

Clonidine and Dexmedetomidine

Meta-analyses of caudal studies in children over 6 months of age, reported prolongation of analgesia with addition of 1–2μg/kg clonidine to local anesthetic for 2.4 (95%CI:2.6–5.5) hours ⁷⁹, 3.98 (95% CI: 2.84–5.13) hours ⁸⁵ and 3.68 (2.65–4.7) hours ⁸⁶. Many studies reported minor sedation following clonidine, which was more severe and associated with cardiovascular side-effects at higher doses (5μg/kg) ⁷⁹. Case reports of side-effects of apnea, oxygen desaturation, and bradycardia have been reported in neonates given doses of caudal clonidine (1.25–2.2 mg/kg) that are tolerated by older children ^128–130. Continuous infusion of epidural clonidine 0.08–0.12 μg/kg/hr produces dose-dependent analgesia when added to local anesthetic infusions ¹³¹, and higher doses of clonidine alone (0.2μg/kg/hr preceded by bolus of 2μg/kg) provide analgesia at rest following abdominal surgery ¹³². When added to intrathecal local anesthetic in neonates relatively large doses of clonidine (up to 2mcg/kg) prolonged analgesia ²⁶. A subsequent observational study with longer follow-up (24 hrs) found over half of the patients were sedated in the immediate postoperative period, and the proportion of neonates developing self-limiting apnea increased postoperatively ²⁷. This dosing represents concentrations up to 5mcg/ml being utilized for both caudal and intrathecal single shot injections and 0.6 to 1mcg/ml for continuous epidural infusion.

The more selective alpha₂-adrenergic agonist dexmedetomidine (1μg/kg) prolonged analgesia when added to caudal bupivacaine, and reduced supplemental analgesic requirements by 1–2 doses of acetaminophen 10mg/kg in the first 24 post-operative hours ¹³³. Similar analgesia was reported when comparing caudal dexmedetomidine and clonidine in children aged 6 months and above ¹³⁴. As there has been limited evaluation of neurotoxicity with this drug¹³⁵, further testing is required before routine clinical use ¹³⁶.

Ketamine

Caudal ketamine has been utilized for perioperative analgesia in children, including neonates and infants ^{84,87,97,98,137}. Dose ranging studies using 0.25–1mg/kg reported 0.5mg/kg as the optimum dose, with increasing side-effects at 1mg/kg ^83,138–140. Recent meta-analyses evaluating addition of ketamine to caudal local anesthetic reported prolongation of analgesia for 2.26 hours (95%CI:1.53–2.98) ⁸⁷ or 5.3 (95%CI:5.45–5.76) hours ⁸⁴. Acute psychomimetic effects were reported in 2 of 7 trials ⁸⁴, but the difference was not statistically significant in the other analysis (OR=1.72, 95%CI:0.69–4.26) ⁸⁷. A reduction in supplementary analgesics was demonstrated in studies utilizing non-opioid analgesics ⁸⁷ or acetaminophen (paracetamol)⁸⁴, but not in studies where peri-operative opioids were required ⁸⁷. Ketamine 0.5–1mg/kg was diluted with 0.5–1.0 mls/kg of local anaesthetic or saline resulting in final concentrations approximating 0.5–1.3mcg/ml ^137,139,141.

Systemically administered S-ketamine has increased potency over the racemic mixture ¹⁰⁰. Dose sparing has not been evident in caudal studies, with s-ketamine utilized in doses of 0.5mg/kg ^142,143 or 1mg/kg ^102,141,144. Ketamine solutions may contain benzethonium chloride ¹⁴⁵, but there is limited information about the injectate preparation in some studies ^112,146,147, while others report using a preservative free solution of racemic ^{106,107,138,139,148,149} or S-ketamine ^97,98,137. In some regions, the number of centers using neuraxial ketamine in children has reduced in recent years ^111,150.

Midazolam

Midazolam is a GABA-A agonist with potential analgesic actions in the spinal cord, but major concerns have been raised about the safety of neuraxial administration in both adult ^151,152 and pediatric practice ¹⁵³. Addition of midazolam 50 μg/kg to caudal local anesthetic prolongs analgesia and increased sedation in children aged 1–12 years ¹⁵⁴. Some reports employ a preservative-free solution ^148,155, but others give no details of the pharmaceutical preparation ^154,156 although one reported using a solution with a pH of 6.2 rather than 3.3–3.9 as used in previous studies ¹⁵⁷. Solutions of 0.1–0.5% midazolam were administered with 0.5–1.0ml/kg of local anaesthetic or saline resulting in final concentrations approximating 50–100 mcg/ml ^{154,155,157,158}.

Neostigmine

Neostigmine produces analgesia following neuraxial administration in adults ^159,160, but the incidence of side-effects has led to its role in pediatric practice being questioned¹⁶¹. Doses of caudal neostigmine ranging from 1 to 4 μg/kg have been administered in children from 5 months of age ^162–166 and prolong analgesia by 9.9hrs (95%CI: 7.8–12.2hrs) but without a clear dose-response relationship ⁸⁶. The relative risk of PONV is significantly increased (RR 1.78, 95%CI: 1.11–2.85] ⁸⁶, with incidences from 30% ¹⁶⁷ and up to 60% with higher doses ¹⁶⁸. Preparations containing methylparaben and propylparaben ^148,169 and preservative free solutions ¹⁷⁰ have been utilized. Prolongation of hyperbaric bupivacaine block has also been demonstrated with intrathecal neostigmine 0.75–1mcg/kg in infants ¹⁷¹. This dosing represents concentrations of 2–4mcg/ml for caudal injections and 10mcg/ml administered intrathecally.

CLINICAL CHOICE OF SPINAL ANALGESIC: SAFETY

For the last two decades, there has been an increasing appreciation that there needs to be a specific intent to define the safety of neuraxially delivered drugs prior to routine clinical use in adults ^172,173. We, and others, have argued that systematic preclinical assessment of potential for spinal toxicity in validated models should be performed before clinical delivery into the neuraxial space of neonates and children ^150,161. Without safety data, it is impossible to confirm a favorable risk-benefit ratio for neuraxial administration, or to compare the relative safety, of this wide range of drugs and preparations, and clinical trials must be undertaken with caution. So significant has become this issue, that several major journals involved in pain and anesthesia have provided specific guidelines on the acceptability of work that employs the off-label neuraxial use of novel agents, indicating that systematic preclinical safety should be available or specific FDA approval gained prior to undertaking the trial ^174–177. In the following sections, we review the information that does exist regarding spinal adjuvant use in human infants; but we emphasize that in and of itself, such information does not qualify the agent being delivered as safe. Often it reflects retrospective series, limited follow-up, and the primary metric of the safety study (i.e. spinal histopathology) cannot be assessed.

Evaluation of risk

Concerns regarding the potential for toxicity following neuraxial analgesic administration have been raised in multiple reviews and editorials with calls for further preclinical testing. “It is essential to undertake extensive animal testing with further evaluation of any neurotoxic effects prior to pediatric use” ⁷⁹. “Before epidural midazolam is routinely used for surgery in children, more extensive testing of its use in animals should be completed”.. and “although the extensive preclinical testing may seem burdensome, the risk-benefit relationship for epidural midazolam justifies the need” ¹⁵³. Although preservatives in preparations of neostigmine ¹⁷⁸ and ketamine ¹⁷⁹ may contribute to potential toxicity, using a preservative-free solution does not guarantee safety. Authors reporting the use of caudal ketamine acknowledge that “as yet, no permanent neurological injury has resulted from single-shot caudal ketamine use but caution is warranted” ⁹⁷, and that conclusive safety studies are required ^84,100. This is particularly important as isolated cases of post-operative neurological injury have been reported in children, and neuraxial analgesia may be implicated in medicolegal claims even if other potential factors (such as peripheral compression neuropathy related to positioning) are subsequently identified ¹⁸⁰.

It was suggested several years ago that performance of neuraxial anesthesia in healthy children required demonstration of a high therapeutic ratio and additional advantages ¹⁸¹. Although complications are rare ³⁵, without information regarding tissue toxicity it is difficult to determine if the drug administered contributes to the risk. Extensive clinical use does not preclude the potential for cases of toxicity ⁷⁹, as seen in adult practice with chloroprocaine ¹⁸² and lidocaine and cauda equina syndrome ¹⁸³. It has also been noted that a single case of neurological injury may be sufficient to change clinical practice, bring a particular technique in general into disrepute, and thus deny many children the benefits of neuraxial analgesia ¹⁶¹. Therefore, further specific data comparing the efficacy and relative safety of currently available and potential new spinal analgesic agents is essential to inform clinical choice. New alternatives should only be used if improved analgesia, combined with an acceptable safety and side-effect profile, can be demonstrated ¹⁶¹. It should be stressed that the neuraxial route of delivery exposes local tissues (meninges, roots, spinal parenchyma) to extraordinary concentrations of agent (mg/mL), which because of local restrictions in redistribution may persist for extended intervals. Accordingly, the specific assessment of the potential toxicity of the agent must be of the highest priority. In the following sections we will review the existing preclinical data related to the safety of spinal anesthetic and analgesic agents in neonatal models.

3. PRECLINICAL MODELS OF NEURAXIAL ANLAGESIA: DEVELOPMENTAL PHARMACODYNAMIC RESPONSES

Neonatal neuraxial delivery models

Intrathecal and epidural delivery techniques

Bolus intrathecal drugs in neonatal and infant rats can be performed with a technique similar to that described in adult mice ¹⁸⁴. The spinal column or pelvic girdle is stabilized by one hand, and percutaneous injection is performed at the level of the cauda equina in the L5/6 interspace (rodents have 6 lumbar vertebrae) with a 30-gauge needle attached to a syringe calibrated to deliver microliter volumes. Correct placement is typically demonstrated by a tail flick on needle insertion. While it is likely that such a response represents contact with a nerve root and is a potential source of pathology ¹⁸⁵, appropriate control studies in neonatal rats have revealed no untoward anatomic pathology related to this technique ¹⁸⁶. Systematic training with the injection of dye and confirmation of spread within the CSF on post-mortem dissection ensures that each experimenter can consistently perform the technique ^184,186. In addition, we recently used in vivo imaging following intrathecal injection of a fluorescent dye to confirm that our technique was reliable and reproducible in rat pups as young as 3 postnatal days that have an average weight around 10 grams ¹⁸⁶.

Intrathecal catheters have been inserted via a lateral thoracic laminectomy in pups as young as P3. An injectate volume of 4μL of methylene blue produces spread from the caudal cervical to lumbar/sacral region ¹⁸⁷, but associated motor deficits limit behavioural analysis to the contralateral limb.

Single shot percutanenous epidural injections can also be performed in rat pups, with correct epidural placement (spread along vertebral segments but lack of staining in CSF) is confirmed by co-injection of Evan’s blue and post-mortem dissection ^188–190.

Distribution of injectate

The distribution of the neuraxially delivered agents must be defined in any preclinical model. The volume must be adequate to deliver agent to the appropriate dermatomes used to evoke pain behavior (e.g. lumbar segments for evaluation of hindlimb withdrawal reftex sensitivity) but insufficient to acutely produce supraspinal redistribution. Recently, we confirmed that segmental spread of intrathecal dye co-varied directly with injectate volume and inversely with age in rat pups ¹⁸⁶. An injectate volume of 0.5 μL/g produced spread across a median of 9, 7 and 5 segments at P3, P10 and P21 respectively. Increasing the volume to 1μL/g increased spread (median number of segments 16 vs 9 at P3, 13 vs 7 at P10). This was confirmed with in vivo imaging, and larger injectate volumes of 1.5 μL/g resulted in fluorescent dye extending into the cisterna magna and supraspinal cisterns ¹⁸⁶.

The extent of epidural spread has also been related to the volume of injectate in several species ^191–193. Similarly in rat pups of different ages, injectate volumes have been based on body weight, and reflect the increasing volume of the elongating spinal canal. In neonatal rat pups, epidural administration of approximately 2 μL/gram of dye ^188–190 produces spread to the mid-thoracic region following low lumbar injection.

Radioactive labeling in the spinal cord has also been used to characterize neuraxial injections. Percutaneous intrathecal injection of 2 μL in P3, or 7 μL of ³[H]-gabazine in P21 rats, produced binding throughout the thoracolumbar cord ¹⁹⁴. Epidural injection of ³[H]morphine at P3, P10 or P21 produced similar levels of binding in the cord, all of which, as expected were much lower than levels seen following systemic administration of the same dose ¹⁸⁹.

An important indirect assessment of correct placement is the observation of an appropriate behavioral response following injection of an analgesic or local anesthetic. While overly large volumes promoting supraspinal redistribution are to be avoided, very small volumes may in fact lead to an inadequate movement of the injectate to the spinal segments regulating the processing of afferent traffic. Accordingly, demonstration of a reliable and dose-dependent change in pain behavior is a critical component of validating dosing volumes in a preclinical model. Neuraxial local anesthetic effects may be assessed by motor and/or sensory changes, and thoracolumbar spread can be assumed by maintenance of adequate respiration, motor block restricted to the hindlimbs and/or lack of a hindlimb withdrawal response to a suprathreshold stimulus ^76,186.

Developmental pharmacodynamic profile of spinal analgesics

We have postulated that evaluation of the relative safety (or toxicity) of different spinal agents is best made in the context of the therapeutic ratio i.e. the dose that produces toxicity or the maximum tolerated dose versus the dose that is required to have a therapeutic analgesic effect ^186,195,196. Accordingly, it is appropriate to consider the utility of neonatal models of neuraxial delivery in defining dose-related analgesic and behavioral effects. Developmentally-regulated changes in the structure and function of nociceptor pathways, and in the expression and distribution of receptors, have a significant impact on analgesic efficacy and dose requirements during postnatal life ¹⁹⁷. Studies in developmental models, particularly the rat pup, allow systematic assessment of a variety of specific nociceptive end-points and the degree of alteration by analgesic agents.

Analgesic efficacy and age-dependent dosing

Increases in the mechanical withdrawal threshold or thermal withdrawal latency threshold of an un-injured hindlimb can be used to evaluate age- and dose-dependent anti-nociceptive analgesic effects. The efficacy of spinal analgesics has also been evaluated by nocisponsive behaviors to local irritants such as formalin ¹⁹⁸ or mustard oil ¹⁹⁹, and also in facilitated hyperalgesic states such as carrageenan-induced inflammation ^188,190,196.

In early life, an enhanced sensitivity to opioids is demonstrated whether given by systemic ²⁰⁰, epidural ^189,201 or intrathecal ^186,202 administration, and lower dose requirements with neuraxial administration confirm selective spinal analgesic effects ¹⁸⁶. Changes in opioid receptor distribution in the dorsal root ganglion and spinal cord are likely to contribute, and may also explain modality specific differences in efficacy against thermal and mechanical stimuli ^{189,202–204}. Lower doses of opioid ^186,201, local anesthetic ¹⁸⁸, NMDA antagonist¹⁹⁶ and alpha2 agonist^190,199,205 reverse injury-induced hyperalgesia and/or increase withdrawal thresholds in neonatal rat pups when the dose is adjusted for weight.

Side effects

Effects unrelated to analgesia may be usefully considered as those which are reversible and those that are irreversible. Side-effects such as sedation, motor impairment, and cardiovascular changes can often limit dose escalation. These dose-dependent effects can be evaluated in laboratory studies and compared with analgesic doses to determine the therapeutic window (difference between dose producing side-effects and the analgesic dose) at different ages. In humans, such side effects may represent: i) a spinal action (e.g. inhibition of the micturition reflex after spinal morphine)²⁰⁶; ii) a direct neuraxial redistribution to the brain (as with the behavioral disruption reported after intrathecal ziconotide) ²⁰⁷; or iii) systemic redistribution of the neuraxial dose after intrathecal delivery (e.g. rapid sedation after intrathecal lipophillic agents such as sufentanil) ²⁰⁸. Side-effect end points may vary with age. Thus, in preclinical models, in addition to lower anti-hyperalgesic dose requirements with epidural dexmedetomidine, the dose that significantly prolonged the righting reflex or reduced heart rate was lower in the youngest animals, resulting in a narrower therapeutic window in early life ^190,199. It should be stressed that these side effects represent adverse events that are related to the physiological and reversible pharmacodynamic profile of the particular competitive agent. Support of function, such as respiration and blood pressure, until drug clearance or reversal will often prevent any further deterioration. These events are important as they limit the useful dose range of the agent that can be practically tested and to which the patient may be safely exposed. This would be defined as the maximum tolerable dose (MTD).

In contrast, drugs at some concentration or dose exposure may exert a direct effect upon cellular function and lead to irreversible changes in cellular viability and thus represent tissue toxicity. Such end points would be, for example, expression of apoptosis or necrosis, frank demyelination, or changes in endothelial cell function. Some of these effects may be manifest by changes in spinally mediated behaviors or physiology, such as seizures, paralysis or anesthesia. On the other hand, where the tissue injury is delimited or where changes are slow and initiate compensatory actions, such effects may not be associated with functional or behavioral changes in the preclinical model. An example of this is the slowly growing, space occupying granuloma ²⁰⁹. Here, the appropriate criteria are the systematic post mortem assessments of target tissues (spinal cord, nerve and DRG). Without this, the absence of negative functional signs can be a false negative as regards tissue toxicity.

PRECLINICAL SPINAL DRUGS: DEVELOPMENTAL TOXICITY

Impact of postnatal age

Preclinical models for assessing intrathecal and epidural drug safety have been established in adult animals ²¹⁰, but there has been little effort until recently to develop models for assessing spinal toxicity throughout the early postnatal period of development. It is crucial that while persistent changes in behavior after neuraxial drug treatment maybe a signature of direct tissue toxicity, absence of such changes cannot be construed as being an absence of toxicity. Such an assertion requires demonstration of absence of neuropathology, e.g. histological signs, increases in apoptosis, and alterations in glial response in exposed tissues. We argue that an important element in considering drugs for neuraxial delivery in human neonates and infants is their appropriate preclinical evaluation. In the following sections we will consider several variables that we believe impact upon the preclinical assessment of developmental toxicity of neuraxially delivered agents.

Activity-dependent neural development

There are well-established critical periods in early postnatal life when the normal development of neuronal circuits is activity-dependent, and alterations in neural activity can produce long-term consequences that are not seen following the same perturbation in the adult ²¹¹. Neural activity promotes synaptic strengthening and network formation; whereas lack of activity and failure to form appropriate synaptic contacts can result in programmed cell death (apoptosis). In contrast to excitotoxic cell death, apoptosis is a normal developmental process for activity-dependent matching of pre and post-synaptic populations and the refinement of neural circuits. However, during these critical periods, exposure to drugs such as general anesthetics that reduce excitation (NMDA antagonists) or enhance inhibition (GABA agonists), may trigger excessive degrees of apoptosis in many brain areas ^11,212–216. The degree and distribution of apoptosis change during the first 2 postnatal weeks, with peak susceptibility in the cortex around P7 ²¹⁷. Prolonged general anesthesia in P7 pups increases apoptosis not only in the brain but also in the spinal cord ^76,218. Changes outlined below also emphasize the significant plasticity of the developing cord. As such, neuraxially administered anesthetics and analgesics which alter neural activity in the cord may also produce specific patterns of toxicity that differ from those seen at older ages.

Developing spinal cord structure and function

During postnatal development there are significant structural and functional changes in nociceptive circuitry in the spinal cord. A-fiber afferent terminals initially project throughout the dorsal horn and only gradually withdraw to deeper laminae over the first 3 postnatal weeks in the rat as C-fiber projections mature ^219,220. The normal postnatal development of A- and C-fiber innervation in the spinal cord is activity-dependent and can be altered by changing input at critical stages ²²¹. Blockade of synaptic activity by a neuraxially administered slow-release NMDA antagonist prevents the structural re-organization of A-fiber terminals and the neonatal pattern of low mechanical withdrawal thresholds and large dorsal horn receptive fields persists into adulthood ²²². The somatotopic organization of primary afferent terminal fields can also be altered by changing neural input during the neonatal period ^223,224. Cell death in the DRG is a normal developmental phenomenon and is balanced by proliferation in early life ²²⁵. However, cell death occurs more rapidly and to a greater extent after sciatic nerve section in neonatal compared with adult animals ²²⁶. Importantly, responses to neonatal injury such as inflammation or surgical injury have been associated with long term functional consequences and an enhanced sensitivity to future injury ^{197,227–230}.

The balance between excitatory and inhibitory activity in the spinal cord changes during the postnatal period^{197,231–233}. Excitatory glutamate receptors (AMPA, NMDA and metabotropic glutamate receptors) are highly expressed and tend to be more widely distributed in the neonatal spinal cord. Developmental changes in subunit expression of the NMDA receptor are associated with changes in channel kinetics and increased calcium influx that further increase excitatory effects^231,232, and may influence the potential for toxicity. GABA inhibition is functional at a cellular level, but there is minimal glycine-mediated inhibition in the neonatal spinal cord ²³⁴ and a delay in the overall maturation of inhibitory networks^{194,233,235,236}, and local GABA-mediated inhibition in the cord is initially dominated by descending excitatory effects^237,238. Ketamine and propofol have been shown to increase cell death and alter dendritic arborization of GABAergic neurons in vitro^239,240 but effects in spinal networks have not been directly evaluated.

4. STANDARDS FOR PRECLINICAL EVALUATION OF EFFICACY AND TOXICITY OF SPINAL ANALGESICS

Characteristics of preclinical safety evaluations

Preclinical safety evaluations by definition employ surrogate models with key characteristics that mirror those of the human condition; in this case, the mammalian neonate during the early post natal phases of development receiving spinal drug exposure in a validated model. As reviewed above, the minimal component to an appropriate assessment of toxicity is the systematic consideration of pathology in the neuraxis as compared to the appropriate neuraxial vehicle control.

Validated model and drug delivery

To date the principal developmental toxicity model employed for neuraxial delivery has been percutaneous delivery in rat pup; but the model (i.e. the animals and the delivery system) must be validated. This implies that the drug delivery has been reliably demonstrated to occur within the intrathecal space (an important issue where the delivery has been percutaneous puncture) and that the injection protocol (needle placement, volume) results in an adequate and reliable distribution of the injectate. As discussed earlier, preliminary studies are required to ensure reliability of the technique in the hands of each investigator, and to avoid confounding effects of dyes in toxicity studies correct placement can be confirmed by measuring a predetermined dose-dependent acute behavioral change (eg. increase in hindlimb withdrawal threshold or motor block). In addition, the model should have the ability and sufficient sensitivity to reveal a profile of toxicity that has been previously described (e.g. apoptosis or demyelination).

It is of fundamental importance that appropriate control groups are included to statistically differentiate between the effects of the interventions and effects of the intervention plus drug. A saline injection group will demonstrate effects related to the technique, needle trauma or volume of injectate. In addition, comparison with a naïve group ensures effects are not related to the brief anesthesia, handling or maternal separation required for the procedure ¹⁸⁶.

Spinal toxicity in adult models has been evaluated following both epidural and intrathecal delivery. Although both intrathecal and epidural delivery have been demonstrated in the neonatal rat, current toxicity models focus on intrathecal delivery. Higher doses or concentrations of epidural drug are frequently required to achieve similar concentrations at target sites within the spinal cord. As such, the worst-case scenario is the intrathecal delivery of an intended epidural agent; not only because of the risk of increased acute side-effects, but also because of the exposure of the cord to an increased dose or concentration of drug. Cases of unrecognized dural puncture and inadvertent total spinal have been reported in large series (2 per 10,633 cases³⁵ and 1 per 10,098¹⁷). In addition, the overall incidence of dural taps has been reported at 0.12% ¹⁶ and 0.1% (CI 95% = 0.05–0.19)¹⁷, and 6 of the ten dural taps in the latter survey were associated with caudals in babies. This further emphasizes the need to establish a safety profile for all neuraxial drugs, whether epidural or intrathecal delivery is planned.

Animal age

The infant rodent is frequently utilized as a model for evaluating the progress of postnatal mammalian development. While direct translation of different developmental ages from rodents to humans, and the specific timing of events after birth, continues to be debated, the sequence of development of sensory and reflex systems in rodents correlate with those of human infants ²⁴¹. Statistical models have been developed to translate development across species ^242,243, but are predominantly based on structural measures, and acknowledge that as peak synaptogenesis is more complex and more prolonged in the human, the model can not account for activity-dependent modification following birth ²⁴⁴. In terms of spinal processing many approximate a P3 rat with a preterm human neonate, P7 with an infant, P21 with an adolescent, and P35 with young adulthood ^197,245,246. Translational developmental models based on correlating behavioral measures support these estimates ²⁴¹. In both humans and rats, locomotor capabilities develop postnatally, with a gradual rostrocaudal pattern of maturation. Rat pups ambulate through use of forelimbs and the upper torso by P3-4, crawling behavior peaks around P7, body weight is fully supported by P12-13, and rearing without foreleg support is achieved by P18 ²⁴¹. Spinal reflexes, which incorporate both sensory and motor development, also show similarities in the sequence of development in the postnatal rat and human infant ^247,248, with gradual maturation from low threshold ^{190,249–251}, large receptive fields ^251,252, poorly directed and generalized responses ^250,251,253 in both rodent and human infant early life. Clear relationships between the intensity of the stimulus and the degree of reflex withdrawal response ^229,254,255 are maintained at all ages, thus facilitating evaluation of the response to injury and/or analgesia.

Vulnerability to apoptosis in the brain coincides with rapid synaptogenesis or the brain growth spurt, which occurs predominantly in the first two postnatal weeks in the rodent, but may extend from mid-gestation to several years after birth in the human infant ²¹⁶. The majority of preclinical studies evaluating general anesthetic effects in the brain have focused on P7 as apoptosis peaks in the cortex at this age, and drug effects are most apparent in regions where spontaneous apoptosis is occurring ^217,256. Spontaneous apoptosis occurs in the postnatal spinal cord, occurs predominantly in the dorsal horn, and peaks at a slightly earlier developmental stage than seen in the cortex with the number of apoptotic cells highest at P2-P5, and decreasing by P8-10 ^{196,257–259}. As peak apoptosis occurs at an earlier age in the spinal cord (P3 rather than P7) than the cortex, the period of susceptibility to pro-apoptotic drugs may be shorter, but prolonged general anesthesia does increase apoptosis in the cord at P7 ^76,218. In addition, as there are ongoing changes in the structure, function and synaptic connectivity of neural networks in the spinal cord throughout the first 3 postnatal weeks ²³², assessment of developmental neuraxial toxicity should include a range of ages. This also addresses the potential uncertainty in the precise parallels between the postnatal development in the human and rodent.

Evaluation and Outcomes

This review will not seek to cover the appropriate histopathology in detail, but experts in the fields of neuropathology will argue that to define the absence of pathology, one must satisfactorily address a number of specific issues and tissue targets.

Blinded assessments

Evaluation must include an analysis that is made independent of knowledge of tissue/animal treatment, with groups that at a minimum include vehicle vs drug treatment cohorts with tissue harvested at predetermined intervals after drug exposure.

Histopathology

Analysis of pathology requires appropriate selection of histopathologial targets and indices.

1. At the minimum, it is reasonable to systematically examine hematoxylin and eosin sections to note necrosis, gliosis, and inflammation. Such examination typically includes spinal cord and meninges, may also include dorsal root ganglia, and evaluation of nerve roots and demyelination is particularly relevant for assessing effects of local anesthetics ^260,261.

2. Evaluation of apoptosis and neuronal cell death is an important additional component in early development. Although a range of potential techniques are available ²⁶², activated caspase-3, an enzyme in the apoptotic cascade which marks neurons progressing beyond the point of commitment to cell death ²⁶³ has been frequently used to identify apoptosis in the brain and also the spinal cord ^{76,186,196,218}. Fluorojade C is an additional marker of neuronal degeneration ²⁶⁴, and we found a pattern of staining that correlated with activated caspase-3 immunohistochemistry ^186,196.

3. Activation of non-neuronal cells by the use of specific astrocyte (GFAP) and microglia (IBA1 or OX42) markers can provide further indicators of altered function and the response to injury.

4. Evaluation of potential nerve injury requires assessment of the state of myelination. Previous work has shown that local anesthetics can produce signs of demyelination of the cauda equina ^265,266. As myelin is in the developing stage up through postnatal day 3, acute effects on myelin may be difficult to assess. Others have focused on apparent changes in the root at later time points, or in the dorsal column which represents the ascending collaterals of large primary afferents ⁷⁶.

5. As mechanisms associated with developmental anesthetic toxicity are further clarified, additional factors requiring evaluation in the developing spinal cord may be identified. As noted earlier, ketamine and propofol have effects on the dendritic tree of cultured cortical and hippocampal neurons.^240,267,268 As changes in dendritic morphology in the spinal cord have been noted in developmental neurological disorders and have a role in synaptic plasticity after nerve injury,^269,270 similar mechanisms may be relevant for developmental toxicity in the spinal cord. Neurotrophic factors and actin depolymerization have been associated with apoptosis in cultured neuronal cells exposed to propofol²⁷¹ and isoflurane,²⁷² but effects in vivo²⁷³ and relevance to analgesic toxicity in the spinal cord has not yet been established.

6. A corollary to this commentary is that evaluation of the potential for spinal toxicity must involve the use of in-vivo animal models. Such models may be complemented by the study of drug effects in ex-vivo or in-vitro models, as has been widely employed to study local anesthetic toxicity. Changes in DRG cell function, or clonal cell viability or ex-vivo nerve exposure^274–277 all provide important approaches to define potential mechanisms. However, as useful as the ex-vivo system is for characterizing local drug effects, care must be taken in extrapolating these results to the intact organism, as they can just as easily provide false positive indications which may not be relevant to in-vivo safety or pathology related to a given drug (see²⁷⁸).

Age at time of exposure

An important issue relates to the developmental age at initial drug exposure. As reviewed above, critical postnatal periods of neural development are represented by the onset of innervation, development of myelination of the long tract and primary afferents, and the time course of spontaneous apoptosis. On this basis, we have argued that appropriate ages in the rat are P3, P7 and P21, with P21 reflecting an animal that has essentially reached a steady state for the end points indicated

Survival time post exposure

Initiation of cell death may begin as early as 6 hours after toxin (drug) exposure, and caspase-3 immunoreactivity may be reduced at later time points as the cell decomposes ²⁶³. In the spinal cord, we found increased apoptosis 6 hours following intrathecal ketamine at P3, and significant increases were maintained at 24 hours ¹⁹⁶. However, glial reactions and evidence of demyelination may not be maximal until a later time point ^{76,196,261,266}. Accordingly, an optimal characterization would include both an early (6–24 hrs) and later interval (7 day) of post treatment recovery. Longer-term effects on functional outcomes must also be considered, and be sufficiently sensitive to detect changes that are related to any observed structural or histological defects. For example, as general anesthetics at P7 increase apoptosis in the hippocampus, long-term effects on learning and memory have been evaluated ¹⁰. Although prolonged general anesthesia increased apoptosis in the spinal cord, motor performance on the Rota-rod at P30 was not altered ^76,218. Whereas local anesthetic toxicity or demyelination may result in changes in motor function, spontaneous apoptosis in the ventral horn occurs mainly before birth ²⁵⁷. Spontaneous apoptosis ^257,259 and increases following intrathecal ketamine at P3 occur predominantly in the dorsal horn, with associated long-term changes in mechanical thresholds for hindlimb withdrawal, and in static but not dynamic parameters of gait ¹⁹⁶. This suggests that alterations in sensory and motor function should be included when evaluating long-term effects of neuraxial drugs.

Drug exposure and dose

To have credibility as a robust assessment paradigm, the drug exposure must occur at neuraxial doses which by the metric of concentration and dose equal or exceed those destined for the human condition. One limitation of percutaneous administration is that effects of dose are limited to single administrations rather than ongoing infusion and chronic exposure. Intrathecal catheterization has been reported in pups as young as P3, but motor deficits and histological damage have been noted ipsilateral to the catheter ¹⁸⁷, thus limiting the utility of this method for assessing toxicity. The use of a single dose runs the evident risk that a drug will be observed to have pathology at a dose which is well beyond any reasonable clinical exposure. Nevertheless the higher the dose examined without pathology the more confident we can feel that the assertion of “no toxicity” is valid ¹⁷³.

Translation of drug exposure and dose

An important question relates to the expression of the dosing, and the translation of dosing in the surrogate to the target species. After systemic delivery the typical metric for dose response is the body mass (eg. mg/kg). However, it is widely appreciated that across large ranges of body weight, a more appropriate metric may be body surface area, particularly when precise dosing is required to maximize the therapeutic response while minimizing the likelihood of unacceptable toxicity (e.g. chemotherapy dosing)²⁷⁹ (Table 3). As BSA has also been shown to correlate across mammalian species with physiological functions (such as metabolic rate, blood volume and renal function), BSA rather than body weight has been used when converting doses across species to humans. The K_m factor (body weight in kg divided by BSA in m²) is often incorporated in formulae for species conversions: e.g. human equivalent dose (HED, mg/kg) = animal dose (mg/kg) × [animal K_m/human K_m]²⁸⁰. Such calculations aim to produce a comparator that generates a proportional plasma level and are important for converting no adverse effect levels (NOAELs) established in preclinical studies to doses used in clinical trials ²⁸¹. However, the FDA also acknowledges that this approach has limited applicability when drugs are administered into anatomical compartments, such as the intrathecal space, where there is little subsequent distribution and where there may be as much as two fold difference in local volume ²⁸². Considering the spinal dose in terms of mg/kg in two adult humans that may differ by a factor of two in body mass may be appropriate for avoiding systemic toxicity or side-effects associated with redistribution or inadvertent injection into vascular structures. However, variability in intrathecal volume is likely to be less, and as toxicity may be more dependent upon the compartmental volume (i.e. cerebrospinal fluid volume and/or its turn over) it is the local concentrations to which the tissue is exposed that is important ²⁸¹. The problem is yet more complicated where one compares across species, and different methods for dose conversion are shown in Table 3. When expressed as age-specific concentrations (total dose in mcg per mcl CSF volume), analgesia is achieved at twice the concentration of morphine and 42 times the concentration of ketamine and clonidine in neonatal pups. The maximum tolerated doses of intrathecal morphine ¹⁸⁶ and clonidine ²⁰⁵ did not produce toxicity in the rat, despite being delivered in concentrations approximating 600 or greater than 10,000 times respectively than concentration required for clinical analgesia. By contrast, intrathecal ketamine¹⁹⁶ produced toxicity at <150 times the clinical concentration. Although these conversions require some assumptions, and are approximate as only limited dose intervals were assessed, they provide comparisons of different agents, and again demonstrate the reduced safety margin of ketamine when compared to morphine and clonidine.

Table 3.

Comparison of intrathecal doses assessed in neonatal humans and rats

expressed in terms of total dose, body mass, body surface area, CSF volume and CSF turn over.
		Analgesic Dose
		Dosing Metric	Human Newborn	Neonatal Rat Pup (3–5 day)
Body Measures
	Neonatal Body weight (Kg)		3500g	12g
	Body surface area (M*2)		Neonate = 0.2m2 20kg=0.8; 60kg adult=1.6	0.0035m2 304 P1=0.0025; P7=0.0045; P21=0.0094 304; Adult 150g=0.025 279
	CSF volume-mL		50 305	0.12 306
	CSF Turn over (mcL/min)		25mls/day 305 17mcl/min 1.05mls/hr	50mcl/day 0.34mcl/min 306 0.02mls/hr
Analgesics			Analgesic dose	Analgesic dose	MTD or toxic dose
	Morphine	Total dose - mcg	24.5 – 70	0.12mcg	36
		mcg/kg	7 25 20 22,307	10 186	3000 (3mg/kg)
		mg/M•2	0.12 – 0.35	0.034	10.3
		mcg/mL CSF Vol	0.5 – 1.4	1 (2x)	300 (600x)
		mcg per mcl/min CSF Turnover	1.44 – 4.1	0.35	106
		Elimination mcg/ml x ml/hr	0.53 – 1.48	0.02	6
	Ketamine	Total dose-mcg	350	36	120
		mcg/kg	IT ~100 § ED: 500 137 – 1000 98	3000 (3mg/kg)	10,000 (10mg/kg)
		mg/M•2	1.75	10.2	3.42
		mcg/mL CSF Vol	7	300 (42x)	1000 (142x)
		mcg per mcl/min CSF Turnover	20.1	105	1000
		Elimination mcg/ml x ml/hr = mcg/hr	7.3	6	20
	Clonidine	Total dose-mcg	3.5	0.36	120
		mcg/kg	1 27	30	10,000mcg/kg 10mg/kg
		mg/M•2	0.017	0.103	3.42
		mcg/mL CSF Vol	0.07	3 (42x)	1000 (14,285x)
		mcg per mcl/min CSF Turnover	0.21	1.05	352
		Elimination mcg/ml x ml/hr = mcg/hr	0.07	0.06	200
	Bupivacaine	Total dose-mcg	1750 – 3500	60
		mg/kg	0.5 288 – 1 24 0.5% solution	3.75 76 *P7: 16 g
		mg/M•2	8.75 – 17.5	13.3
		mcg/mLCSF Vol	35 – 70	500 (7–14x)
		Mcg per mcl/min CSF Turnover	102 – 205	176
		Elimination mcg/ml x ml/hr = mcg/hr	37 – 73	10

Reference shown in parentheses

^§adult analgesic doses: 0.2–0.5mg/kg epidural analgesia; 0.1mg/kg intrathecal. As 0.5mg/kg caudal epidural also analgesic in neonate, assume 20% dose effective intrathecal

The therapeutic ratio of toxic to analgesic dose: a way forward?

The relative efficacy and safety of different treatments, and the potential benefits and risks for individual patients, are essential for choosing the most appropriate drug in clinical practice. Safety studies frequently appreciate that every agent examined neuraxially will at some point display pathology. The issue is that the drug must have a therapeutic dose that is lower than the dose that produces untoward effects upon behavior or exerts direct tissue toxicity. Ideally, this therapeutic window is wide, but depending on the desired outcome a narrower margin may be tolerated. For example, chemotherapeutic agents produce significant side-effects and toxicity, but the potential benefit for the patient is deemed to outweigh this risk. Similarly, despite concerns about pro-apoptotic effects of general anesthetics, it is clearly not appropriate to withhold anesthesia for neonates requiring surgery. However, if several agents produce a similar therapeutic end point (e.g. analgesia) what algorithm might we use to select the one least likely to have a deleterious action? One strategy is to define the therapeutic ratio of the several agents under identical conditions. In this case, one notes the quotient of the minimum dose without tissue toxicity and the minimum dose required to produce a therapeutic effect of the intrathecally delivered agent. In recent studies we showed that the therapeutic ratio in early life was > 300 for morphine and clonidine, but < 1 for ketamine as increased apoptosis occurred in the same dose range as analgesia ^186,196,205. While the ratio can vary for different reasons across end points and laboratories, we would argue that in a given assessment paradigm, if two drugs have similar analgesic efficacy, but differ in their therapeutic ratio, the agent with the higher therapeutic ratio will be preferred, all other things being equal.

This particular strategy provides a rationale in the current environment to minimize the potential complications secondary to direct tissue toxicity, particularly where old drugs are being given by a new route (e.g. neuraxial) or where new drugs or preparations are being considered for neuraxial use. As noted above, with ongoing clinical use, it can become apparent that even commonly employed agents may lead to pathology; as seen following intrathecal infusion of local anesthetics ²⁸³ and chronic intrathecal morphine ^284,285.

CONCLUSION

We acknowledge that neuraxial anesthesia is an important component of perioperative pain management in children of all ages, and particularly in neonates and infants as inadequately controlled pain in early development may also have adverse long-term effects ¹⁹⁷. Our aim is not to discourage use of neuraxial anesthesia, but rather to encourage use of agents with demonstrated efficacy and the widest possible safety margin. Clinical studies are well suited to assessing tolerability and efficacy, but cannot reliably confirm safety and an absence of morphological effects¹³⁶. Therefore, we complete this overview of neonatal neuraxial analgesic utilization by emphasizing four points.

First, we believe it is evident that the potential for spinal drug toxicity may present a greater problem in early life because of the dynamic properties intrinsic to neuraxial development.

Secondly, given the above issues, we believe that advances in this area require systematic preclinical assessments of the comparative safety of candidate agents with attention being paid to the therapeutic ratio of the neuraxially delivered agent, the developmental time of exposure to the agent, and assessment of neuropathology (apoptosis, myelination, gliosis and dendritic morphology) and long term functional outcomes. Further, the research must recognize that the critical periods of development that occur (e.g synaptogenesis, myelination and apotosis) differ for brain and spinal cord. Of equal importance, as the algorithm relating rodent and human neonatal development cannot be precisely matched, preclinical safety evaluations must review a range of developmental ages in their respective models.

Thirdly, there is a need for a greater appreciation by institutional review boards regulating clinical trials, and by editors and reviewers of scientific publications, of the issues of potential toxicity and the degree to which the clinician-investigator has adequately addressed these concerns.

Finally, we must entertain a high index of suspicion of potential toxicity when drugs are administered neuraxially. As children are rarely subject to detailed assessment after day-stay surgery, there is the potential to under-estimate the rate of complications⁶⁷. This is particularly important in neonates and infants, who may not only be more susceptible to perturbations in neural development, but who are also unable to report sensory symptoms and as they are not walking, subtle motor deficits may be missed. We agree with others, that more thorough follow-up of children following neuraxial analgesia is required⁶⁸, with longer-term epidemiological studies to establish clinical safety²⁸⁶. Integrating preclinical and clinical data has also been the focus of studies evaluating adverse neurodevelopmental outcomes following general anesthetic exposure in early life. In this situation, the clinical benefits of diagnostic investigations and surgery with adequate anesthesia outweigh the risks identified in laboratory studies, and although modifications in practice have been suggested²⁸⁷, current data do not support significant changes in clinical practice or provide clear evidence of a better alternative^11,215. However, when considering the choice of spinal analgesic adjuvants, many provide similar analgesia but not all have undergone systematic evaluations of spinal toxicity, and changing practice to include only agents with the widest demonstrable safety margin can be achieved without compromising clinical care. It is essential to ensure that every step is taken to evaluate both the benefits and the safety of new and existing spinal drugs, prior to routine clinical use, to minimize the risk of an unexpected and untoward outcome.

Figure 1.

Methods used in the evaluation of anesthetic and analgesic toxicity in the developing spinal cord. Preclinical studies in the postnatal rat have used a range of functional/behavioral and histopathological tests to evaluate spinal cord responses to analgesic and anesthetic drugs.

1) Experimental groups include a range of intrathecal drugs. Appropriate control groups include: intrathecal saline and naïve animals. Positive control groups are required to ensure adequate sensitivity of the tissue analysis when evaluating drugs with no or limited toxicity.

2) A measurable behavioral response is required to confirm correct intrathecal placement. Analgesic response curves are schematic examples of dose-dependent reversal of hyperalgesia and of age- and dose-dependent changes in mechanical withdrawal threshold.

3) Neuronal apoptosis and acute histopathological change can be evaluated by a range of methods. Images are representative examples of spinal cord sections stained for: activated caspase-3 immunoreactivity with positive cells identified by brown DAB staining; and Fluor-Jade C staining with apoptotic or dying neurons identified by green immunofluourescence.

4) Histopathological responses to drug exposure, and demyelination following local anesthetics are evaluated at later time points. Images represent spinal cord sections stained with: hematoxylin and eosin to assess histopathology and apoptotic cell counts; Iba1 immunoreactivity to evaluate microglial reactivity (green); and glial fibrillary acidic protein (GFAP, red) to evaluate astrocytes.

5) Measurement of spinal reflex sensitivity, analysis of gait, and length of time on a rotarod have been used to evaluate spinal cord functional outcomes in later life.

Legend: P, postnatal day; I.T. intrathecal; H&E, hematoxylin and eosin