Somatosensory Evoked Potentials in Infants and Children

Examination of the sensory pathways is particularly difficult in infants and young children, and the recording of somatosensory evoked potentials (SEPs) is therefore important as a means of providing objective information about the functional integrity of the somatosensory pathways. As in adults, SEPs are helpful in diagnosing certain disorders, localizing lesions, following the response to therapies, and monitoring the integrity of the sensory pathways during procedures that may put them at especial risk. In the pediatric population, they can also provide information on the development and maturity of the sensory pathways.

Cortical SEPs can be recorded from at least the 27th week of estimated gestational age (i.e., the age in weeks of the infant at birth as calculated from the mother’s last menstrual period; EGA). Although the responses at this time may be difficult to record, are sometimes absent, show marked individual variation, and may differ from those recorded later in life, they show that the somatosensory pathways can conduct impulses from the periphery to the cerebral cortex. In fact, Taylor and colleagues were able to record SEPs in all of 22 preterm infants ranging between 27 and 32 weeks EGA. The waveform consisted of triphasic positive and negative components with latencies that decreased rapidly over the preterm period. There appeared to be no differences in the maturational changes as a function of gestational age at birth.

The generators of the various components of the SEPs elicited by stimulations of limb nerves were discussed in Chapter 26 , to which the reader is referred. Guidelines for recording SEPs are available.

The various components of SEPs are labeled by their polarity and expected latency. Given the differences in length and maturity of the nervous system that exist between adults and children, the latencies of the individual components—and thus their designations—also differ. Table 27-1 shows selected componenets of the median- and tibial-derived SEPs, with their usual latencies for adults and young children.

Table 27-1

Selected Components of the Median and Tibial Somatosensory Evoked Potential in Children and Adults

			Children
	Origin	Adults	Term infants *	1–8 yrs
Median SEPs	Brachial plexus	N9/P9	–	N7/P7
	Medial lemniscus	P13–P14	P10	P9–10
	Cortical area 3b	N20	N27–30	N16
	Cortical motor area 4	P22	P36–39	P20
Tibial SEPs	Sacral plexus	P18	–	P11
	Lumbar potential	N22	N8	N14
	Gracile nucleus	P31	–	P21
	Somatosensory cortex	P37	Variable (P33)	P28

* The most pronounced maturational changes occur in the first 6 weeks of life. Methodological differences make it difficult to compare values obtained by different authors for infants. The values shown here, derived from the literature, are provided simply as a guide.

Methodology

Techniques for recording median SEPs in infants and children are well described. A stimulus rate of 4.1 Hz, stimulus intensity just above motor threshold, and sweep time of 50 or 100 msec are used widely. In normal infants, a bandpass of 10 or 30 to 3,000 Hz, with 100 k gain, commonly is preferred because it allows better differentiation of the early positive and negative cerebral components in the recordings. A bandpass of 5 to 1,500 Hz is helpful in some cases, however, to facilitate recording of late slow waves which may be the first indication of the presence of cortical potentials. For recording the median SEP, a common montage is between the ipsilateral and contralateral Erb’s point (channel 1); the fifth cervical spine and contralateral Erb’s point (channel 2); midway between the ipsilateral C3/4 and P3/4 (CP placement; also known as C3′ or C4′, located behind the C3 or C4 placements in the international 10–20 system), and the contralateral Erb’s point (channel 3); and contralateral and ipsilateral CP placements (channel 4). In infants, recordings are sometimes made at a site 1 cm above the axilla rather than at the ipsilateral Erb’s point to obtain more reliable recording of the Erb’s point potential. Figure 27-1 shows median SEPs in a child of 7 years and another aged 23 months.

For the tibial nerve, recordings are made from over the popliteal fossa (channel 1) and the L1 spine (channel 2) referenced to the sixth thoracic spine, and over the C7 spine (channel 3) and CPz (channel 4) referenced to Fz. Methods are similar to those used in adults. An example of a tibial SEP recorded in an 8- year-old boy is provided in Figure 27-2 . Normative data corrected for height are helpful for evaluating latency values. The scalp-recorded tibial SEP is not seen in infants younger than 31 weeks conceptual age (CA; i.e., EGA plus the legal age), and its presence in older infants is variable ; not all components are recorded, even at term.

Recordings ideally should be made with the patient awake, as cortical components are affected by sleep, but this is not always possible. Sedation should be avoided if possible in infants younger than 3 months of age, but may be necessary in older infants and children. Sedation or anesthesia may attenuate the cortical SEP components and affect component latencies.

Effects of maturation and related factors

Maturation of Structures in the Nervous System

It is difficult to determine the effects of maturation on the somatosensory system because a number of different factors are involved including—but not limited to—change in the length of the sensory pathway and the myelination of different parts of the sensory pathway at different times.

As the human infant develops, nerve fibers myelinate and the diameter and length of peripheral nerves increase. In newborns, the conduction velocity of peripheral nerves increases linearly with CA from approximately 20 m/sec at 33 weeks CA to 33 m/sec at 44 weeks CA. The conduction velocity of peripheral nerves continues to increase with age during early childhood, with maximal values attained at 18 to 27 postnatal months for the sural nerve.

Maturation of the spinal cord occurs later than that of peripheral nerves. At 26 weeks CA, the spinal canal finishes at the level of the fourth lumbar vertebra, whereas by term it is at about the second lumbar vertebra. As summarized by Gilmore, myelin first appears in the fasciculus cuneatus at 14 weeks CA and in the fasciculus gracilis at 22 to 24 weeks CA in humans. With maturation, fiber diameter and conduction velocity increase, exceeding the effects of elongation of the spinal cord in early infancy. Cracco and co-workers used surface distance measurements to calculate conduction velocities over different segments of the spinal cord in infants and children. Conduction velocity was faster in rostral than in more caudal segments or peripheral nerve, and increased progressively with age. It came to be within the adult range in peripheral nerve by 3 years of age, and in the spinal cord by 5 years. In children, the increase in conduction velocity seems to parallel the increase in length of the cord.

The more rostral parts of the somatosensory system myelinate over a wide range of times, probably accounting, at least in part, for the different times at which scalp-recorded potentials or their various components appear in preterm newborns. In the medial lemniscus, myelination begins at approximately 23 to 25 weeks CA in a caudal-rostral order, being completed probably at about 12 months post-term. The thalamus is known to have no or little myelin at approximately 32 weeks CA, but myelination occurs over the ensuing weeks so that by term, the thalamus usually is myelinated. The somesthetic radiations are not myelinated completely until approximately 12 to 18 months post-term.

In infants, the absence of SEP components may simply reflect the fact that maturation has not proceeded sufficiently for development of either the relevant neural generators or the pathways conducting the potentials of interest. With regard to the median SEP, the cortical potential was recorded consistently after 6 post-term weeks in one study, and after 8 weeks in another. If infants without cerebral potentials at birth are retested at 2 to 3 months, most (80 percent) will have cerebral potentials. Maturational changes occur throughout infancy but are fastest during the first 3 weeks of life. In the preterm and term periods, maturation is reflected mainly as shortening of the latencies and increase in the amplitude of especially the early responses of the SEP.

Fifty-two sets of cortical SEPs were recorded from 23 normal children between the ages of 1 day and 13 weeks with median nerve stimulation by George and Taylor, who found that a component corresponding to the adult P22 became the major component by 2 to 3 weeks of age. Gibson and associates studied the median SEPs in 40 healthy term infants. Identifiable potentials were recorded over the cervical spine in all of them and from the scalp in at least some runs in most (39) infants. The cervical response showed little variation and consisted of a clear negative wave with up to three peaks, the largest having a mean latency of 10.2 ± 0.7 msec, followed by a positive deflection. The cortical response was variable in form and latency between infants and in the same infant at different times. Its morphology could be symmetric, asymmetric, plateau- or M-shaped, of increasing complexity. In some instances the response was absent or ambiguous but usually could be resolved by altering the stimulus frequency or intensity. In the whole group, the mean latency for the potential corresponding to the N20 component was 30.0 ± 6.8 msec, and for the central conduction time was 19.8 ± 6.5 msec. Differences between the different cortical waveforms suggested that the M-shaped form was the most mature response.

With regard to the tibial SEP, there is also some variability in the findings. In 26 babies, aged between 1 day and 3 months, Georgesco and co-workers found no reproducible tibial SEPs over the scalp in 6 instances but, in the other 20 infants, an electropositive component (corresponding to the P37) was found that showed no correlation in its scalp distribution or latency with the baby’s size, age, or physiologic state, or the intensity of stimulation. In another study, cortical and spinal SEPs were recorded after median and tibial nerve stimulation in healthy newborns. Spinal SEPs were obtained and recorded readily in all but one neonate after stimulation of both nerves. Cortical SEPs were recorded more frequently after median nerve (87 percent) than tibial nerve stimulation (73 percent) but the shape of cortical SEPs obtained after tibial stimulation was less variable.

In a study of spinal, subcortical, and short-latency cortical SEPs following median or tibial nerve stimulation in 100 sedated or nonsedated children aged between 4 weeks and 13 years, the morphology of the responses was similar to adults, but the initial cortical components of the median SEP showed maturational changes in both interpeak latencies and morphology. The negative peak latencies recorded over Erb’s point (N9 equivalent) and the second cervical vertebra (N13 equivalent) following median nerve stimulation, and over the lumbothoracic junction following tibial nerve stimulation were related directly to patient age and limb length. There was no correlation between age and the latencies of either the initial negativity or positivity of the cortical SEPs after median and tibial nerve stimulation, respectively. Central somatosensory conduction time declined slowly during the first decade and reached adult values after 8 years of age.

The presence, morphology, and latency of the SEP, especially of the cortical component, may reflect the wide time-range over which myelin appears at different sites in the developing central nervous system (CNS).

Age and Height Effects on SEP Latencies

Changes in the length of the pathways mediating the SEPs—as well as maturational changes—influence SEP latency. The latency of the N20 (N16 in the child) component of the median SEP decreases for the first 4 or 5 years of life and then starts to increase with increase in body and arm length. The latencies of the Erb’s point potential and N13 (N9 in the child) are similar in newborns and older infants, probably remaining stable or decreasing slightly until 4 to 5 years. They then start to increase, reaching adult values in the teenage years.

The central conduction time, calculated from the median SEP as the difference in latency between N20 (N16 in children) and N13 is correlated inversely with age, diminishing as the infant grows. It declined slowly during the first decade, reaching adult values after 8 years of age in one study.

In the SEP elicited by fibular (peroneal) nerve stimulation, the conduction velocity along the peripheral nerve reaches the adult range by 3 years of age, whereas that recorded between the lumbar and cervical spinal cord does not do so until the fifth year.

As regards the tibial SEP, a study of preterm infants revealed that the latencies of the potentials recorded over the lumbar and cervical spine following bilateral simultaneous stimulation decrease with increasing CA, probably because conduction velocities are increasing faster (with myelination) than the length of the conducting pathway. By contrast, latency of the popliteal fossa potential is constant and unrelated to age or length, presumably because leg length and conduction velocity increase at similar rates. The latency of the scalp-recorded equivalent of the P37 in adults (P55) shows marked variation, is independent of most factors, and probably reflects variable rates of cerebral myelination and neuronogenesis, varying states of alertness, and possibly subclinical encephalopathies. In children aged between 1 and 8 years, latencies of the early components of the tibial SEP increase with increasing age or height, whereas that of the scalp-recorded P28 potential (the P37 of adults) correlates only modestly with height and reflects asynchronous maturation of elongating polysynaptic pathways.

Effects of Sleep

Tactile stimulation of the fingertips elicits responses over the scalp that are similar in waveform and latency to the responses evoked by electrical stimulation in term neonates. An initial negative potential (N1) with a latency of 31 to 34 msec is followed by one or two positive deflections, P1 and P2, within about 300 msec from the stimulation. Pihko and associates found that N1 was present clearly only in active sleep and only in a few subjects; both positive potentials were depressed in active sleep.

The effect of sleep on the short-latency median SEP is variable in infants and children. The findings in wakefulness and rapid eye movement (REM) sleep are similar. Hashimoto and associates examined electrically elicited median SEPs in 83 children aged between 1 month and 16 years. Components comparable to the adult N20 remained stable during sleep, whereas P22 demonstrated an increase in peak latency during slow-wave sleep but not during REM sleep. In another study on the effect of sleep on the median SEP in 20 healthy children aged 20 days to 3 years, non-REM sleep influenced the latency and the duration of SEPs, especially in children younger than 1 year. In adults, non-REM sleep may prolong SEP latencies.

The effect of sleep on the cortical components of the tibial SEP of infants and children is even more marked and complicated. In some preterm infants, natural sleep actually enhances the appearance of the later cortical potentials.

Clinical applications

Prognosis in Coma

Clinical assessments using the Glasgow Coma Scale may be limited for technical reasons, as when it is not possible to examine motor function because of the effects of muscle relaxants or sedatives. Having the means of assessing the integrity of the patient’s nervous system electrophysiologically as well as clinically is therefore important in guiding prognostication.

The value of the median-derived SEP for prognosis has been examined in comatose infants and children. Absence of cortical components of the median SEP with preservation of the brainstem auditory evoked potential (BAEP) correlated with loss of cortical function and preservation of brainstem function in five children with hypoxic insults, as might have been anticipated. In children with brainstem pathology, however, the recording of both BAEPs and median SEPs may provide more information than does either test alone, but does not necessarily provide a better prognostic guide. SEPs are more useful than visual evoked potentials (VEPs) in evaluating prognosis.

De Meirleir and Taylor reported the SEP findings in children who were comatose from a variety of causes. Those with bilaterally absent cortical SEPs died or were left with a severe spastic quadriplegia; those with unilaterally abnormal cortical SEPs developed a residual hemiparesis; and those with normal or only mildly abnormal SEPs that normalized within a few days had normal outcomes.

Goodwin and associates also found that SEPs and BAEPs are useful in predicting outcome in comatose children. They studied 41 children with a variety of disorders admitted to a pediatric intensive care unit with a Glasgow Coma Scale score of less than 8. Survivor outcome was determined at discharge and 1 to 3 years later. There were no false pessimistic predictions (i.e., predictions of a bad outcome when a good outcome actually occurred), and two false optimistic predictions in this series. A comprehensive literature review of coma outcome prediction, using multimodality evoked potential recordings, revealed 20 series with 982 additional patients in whom the predictive errors of false pessimism and false optimism could be determined. Five cases of false pessimism (which declined to three if neonates were excluded) and 99 cases of false optimism were identified in the 982 additional patients.

Beca and co-workers have studied this issue and compared the SEP as a predictor of outcome in acute, severely brain-injured infants and children with prediction using the motor component of the Glasgow Coma Scale. They found that a normal SEP had a positive predictive value (PPV) for favorable outcome (using the Glasgow Outcome Scale) of 93 percent. Absent SEPs were predictive (92 percent) of an unfavorable outcome (either severe disability, survival in a persistent vegetative state, or death). Although an absent motor response to a painful stimulus also had a PPV of 100 percent for an unfavorable outcome, in 23 percent of patients the response could not be evaluated because of the effects of muscle relaxants or sedatives.

In a prospective cohort of 57 consecutive children who were mechanically ventilated for hypoxic-ischemic encephalopathy in a tertiary pediatric intensive care unit, 42 had impaired consciousness or remained in coma after 24 hours. Among this group, an initial cardiopulmonary resuscitation duration longer than 10 minutes and a Glasgow Coma Scale score of less than 5 at 24 hours after admission were associated with an unfavorable outcome (PPV, 91 and 100 percent). A discontinuous electroencephalogram (EEG) and the presence of epileptiform discharges were associated with an unfavorable outcome (PPV, 100 percent for the two criteria). The bilateral absence of the N20 wave on short-latency SEPs had a PPV for unfavorable outcome of 100 percent. Clinical assessment combined with EEG and SEPs thus allows an early prediction of the prognosis of children.

Clinical assessment may be as predictive or have a higher PPV for outcome than electrophysiologic studies, but is often hampered by sedation or paralysis in patients in the intensive care unit. SEPs and other evoked potential studies thus remain a valuable tool for those caring for comatose children.

Reye Syndrome

In children with Reye syndrome, longitudinal SEPs may be helpful in prognostication. Patient survival in one study was correlated with early recovery of initially absent short-latency (less than 50 msec) cortical components of the median SEP; lack of their recovery was associated with death. Progressive recovery of SEP components later than 100 msec was associated with satisfactory clinical recovery; when these components failed to recover, patients were left with a residual neuropsychologic deficit.

Assessment of Term and Preterm Newborns

The importance of early afferent activity for normal brain development is widely known and indicates the importance of assessing brain responses to afferent stimuli in patients in the neonatal intensive care unit. Evoked potentials are useful clinically in preterm neonates because the neurologic evaluation of these infants is often unreliable. Prediction of outcome for graduates of neonatal intensive care units is helpful in identifying patients requiring early interventions and in guiding how families are counseled.

SEPs, alone or in combination with other electrophysiologic studies, can be helpful for determining prognosis in high-risk newborns. In one study, almost one-third of newborns at high risk for neurologic or developmental sequelae had abnormal SEPs with delayed or absent scalp-recorded potentials or a prolonged central conduction time. Patients were tested again at 2 and 6 months of age, and then neurodevelopmental outcome was determined at 1 year. SEP abnormalities that persisted or worsened correlated with severe neurologic impairment, whereas an abnormal SEP that improved or normalized in infancy was associated with only mild to moderate neurologic sequelae. In a subsequent prospective study of the long-term predictive value of SEPs, BAEPs, and the Einstein Neonatal Neurobehavioral Assessment Scale, 78 high-risk newborns and 28 healthy controls were assessed by these means in the newborn period. At 8 to 9 years of age, 42 subjects and 13 controls were re-evaluated for developmental progress using a range of psychologic, sensorimotor, and neurologic measures. The SEP was most accurate at predicting outcome at school age, with high specificity (83 to 100 percent) across all domains tested and good sensitivity (80 to 100 percent) for intellectual performance and sensorimotor abilities. The BAEP was limited by false negatives, and the neurobehavioral assessment by many false positives. Thus, associations between neonatal SEPs and developmental sequelae remain significant at school age.

Taylor and her co-investigators examined median SEPs and VEPs in asphyxiated term infants who were followed for 18 to 24 months. Electrophysiologic studies were performed in the first 3 days of life, during the first week, and at follow-up visits. The SEPs had high sensitivity (96 percent) and negative predictive power (97 percent); normal SEPs virtually guaranteed normal outcome. VEPs had both a specificity and positive predictive power of 100 percent; abnormal VEPs guaranteed abnormal outcome. Both together had a higher predictive power than either alone. Others have reported similar findings at both intermediate and long-term follow-up.

Median SEPs provide reliable data that can help to predict neurologic outcome in preterm infants at risk of periventricular leukomalacia, and the tibial SEP may be even better but is more difficult to obtain.

Predicting Cerebral Palsy

The above studies bear on the issue of predicting severe motor impairment at an early stage. Infants with prolonged or absent cortical SEPs all demonstrated fixed handicaps after the first year of life, indicating that the SEP is a valuable early indicator of severe motor impairment, although there was a rather high false-negative rate, probably because only the afferent pathway was tested electrophysiologically. In another study, cortical SEPs were always present in the perinatal period in those surviving without major neurologic disability, while they were absent bilaterally in all but one patient with a subsequent diagnosis of cerebral palsy.

In a more detailed study, it was found that SEPs recorded early in the course of the life of a preterm infant predict long-term neurodevelopmental outcome. Thus, unilateral, median SEPs were recorded twice in the first 3 weeks of life in a group of preterm infants. Bilateral SEP abnormalities were associated with the presence of periventricular leukomalacia on head ultrasound and were predictive of later cerebral palsy. False-positive results were frequent. A normal SEP, even when there was periventricular echogenicity on head ultrasound, was associated with a normal outcome in all but one instance. The SEPs were less accurate than the findings of periventricular leukomalacia on ultrasound in the prediction of later cerebral palsy. SEPs done in the first 3 weeks of life therefore may provide additional prognostic information, particularly when the test is normal. Abnormal SEPs in this period must be interpreted cautiously.

Neurodegenerative Disorders

Abnormal spinal and cortical SEP components have been reported in a variety of neurodegenerative disorders. Severe abnormalities have been described in children with polioencephalopathies (primarily disorders affecting gray matter) and leukodystrophies. In children with leukodystrophy (metachromatic leukodystrophy, Pelizaeus–Merzbacher disease ( Figs. 27-3 and 27-4 ), Krabbe disease, adrenoleukodystrophy, Alexander disease, Canavan disease, and multiple sulphatase deficiency), the SEPs were abnormal in all instances. Cortical SEPs were absent in 16 and abnormal in 5 who were in the earlier stages of their disease. Cervical SEPs were within normal limits, except for the patients with Krabbe disease and metachromatic leukodystrophy, who showed peripheral slowing.