AUDITORY BRAIN STEM POTENTIALS RE CORDED AT DIFFERENT SCALP LOCATIONS IN NEONATES AND ADULTS

The auditory evoked brain stem potential was recorded in 14 normal full-term infants and nin~ normal-hearing adults. Silver-silver chloride electrodes were placed at nasion, forehead, vertex, each mastoid over the bony prominence, and the seventh cervical vertebra (noncephalic reference) in order to study the scalp distribution of the auditory brain stem response. Large differences in the scalp distribu tion between the newborn and adult populations were observed. At the ipsilateral mastoid, an x wave occurring at approximately 2 ms and a y wave occurring at approximately 3.3 ms were identified in the adult; this contrasts to a y wave at approximately 3.7 ms in the neonate. It appears that there are either separate generators for some of the components in the adult versus the neonate, and/or as the nervous system matures, mye!inization occurs with a concomitant change in the scalp distribution of the auditory brain stem potentials.


INTRODUCTION
A sequence of potentials is recorded from the scalp in both humans and animals within the first 10 ms following an auditory stimulus. These potentials are the far-field reflection of activation of the brain stem auditory nuclei and pathways. 1 • 1 These auditory brain stem responses (ABR) are clinically useful in assessing cochlear function and in diagnosing neurological disorders affecting the brain stem .1. 2 Although the general anatomical origins of these potentials seem well established, their precise neural generators are not well understood. Detailed information on the topographical scalp distribution of the ABR could help to understand the generators of the ABR.
The waveform morphology of the ABR varies as a function of recording sites over the human scalp. 1 • 5 · 6 • 8 Amplitudes, latencies, and polarities of the peak responses vary according to elecrode placement. In addition, differences in the amplitudes and latencies of various peaks in the ABR have been identified and vary as a function of contralateral versus ipsilateral stimulus presentation. u Further latency disparities are observed and may be emphasized by the use of differential recordings. 1 · 1° The interpretation of the mechanism by which these shifts occur is complex since, in differential recordings of the ABR, both electrode sites are active, and positive (usually the vertex) and negative are a matter of definition.
One of the more interesting aspects is the development and maturation of the ABR potentials. Although many investigations describe the latency and amplitude changes in the maturing infant for standard electrode arrays, eg, Cz-Mi, Cz-Mc, etc, 11 there is neither a description nor a comparison of scalp distribution changes with maturation in the neonate. This investigation compares the scalp distribution of the ABR in the neonate and adult. In · addition, we compared the referential with differential recordings to define the lateralization of the ABR generators.

METHODS
Fourteen normal full-term neonates w ith no ~istory of prenatal or perinatal complications and nine normal-hear ing adults were tested , one man and eigh t women ranging in age from 21 to 37 years. The neonates were studied during light sleep following a normal feeding period. The adults were tested d uring a quiet awake state in a supine position with eyes closed.
The acoustic stimulus used to evoke the ABR consisted of condensation clicks generated by a 0 .1-ms pulse applied mon aurally to a TDH-39 earphone in a MX41AR cushion. Presentations were made at 11 . J clicksls at a 90-dB peak equivalent SPL .
The amplifier gain used to record the ABR was 80,000. T he amplified response was filtered bo~h from 5 to 3,000 Hz an d from 150 to 3,000 Hz (3 dB down points, 6 dB/octave). Studies by Scherg' indicate that a high pass filter of 5 Hz for the ABR does not introduce distortion of the response. A set of duplicate averages were completed at each filtrr setting for 2,048 stimu lus trials comprising each average. Sweep time was 20 ms, with a digitizing rate/address of 25 kHz. Four channels were averaged simultaneously and stored on a floppy disk . Paired t tests were used to evaluate latency differences among electrode locations. For the purposes of this paper a level of significance of p< .02 is used. This level was chosen as a more conservative approach due to the large number oft tests which were used. However, specific levels are given as a matter of form to provide the reader with additional information.

RESULTS
There are significant differences in the scalp distribution of the ABR between the newborn and adult populations.
It should be noted that in some instances there is an apparent discrepancy between mean latencies reported in the Tables and Figures of this study.  The mean latencies in the Tables reflect equal  weighting, eg, each individual latency, whereas, the mean latencies shown in the Figures may reflect an individual bias, eg, since the waveforms were digitally summed to obtain the grand average waveforms, unequal weighting occurs due to individual differences in waveform morphology.

ADULTS
The grand average of the ABR for the adults is illustrated in Fig 1, and a single case recording is illustrated in Fig 2. Referential Recordings. Figure 1 shows the ABR components constructed from monopolar recordings made at Cz. These are labeled at the peaks by Roman numerals and by an n at the following Hz-filter settings. However, we chose the wide band pass (5-3,000 Hz) for all measures of amplitude and latency of the ABR to minimize phase shifts and waveform distortion provided by our analog high pass filters.
The noncephalic electrode at Cvn is considered referential since recordings between that site and a second noncephalic electrode placed on the sacrum did not demonstrate replicable components in the latency domain of the auditory brain stem potentials.
The waveform morphology at Mi, particularly in the first few milliseconds, differs in polarity and form from the other recording sites (Figs 1 and 2). The first negative deflection at Mi corresponds to a positive deflection at Cz and Na in the grand average adult recording in Fig 1, and flat, or perhaps slightly negative, at Mc. The individual example from an adult subject (Fig 2) shows Cz, Mc , and Na having a positive deflection. A large amplitude positive wave (x) between wave I and In is seen at Mi in Figs 1 and 2. For the individual recording, the x wave is not seen at Cz, Mc, or Na. However, in the grand average recording (Fig 1) it may be present at Mc. The second negative wave at Mi corresponds in latency to wave In at the other recording sites. Wave II is positive at all recording sites, as is wave Iln negative at all recording sites for the grand average (Fig 1). This is likewise generally true· for the individual average ( Fig  2). A positive peak, y, occurs between waves Un and III at Mi and is not seen at other recording locations. Waves III and Hin at Mi show both a polarity reversal and latency shift compared to the other electrode locations. The remaining waveforms (waves IV through Vn) are similar at all electrode locations.
Mean latencies of the ABR at the four recording sites are shown in Table 1. In evaluating the latency differences between the ABR components for the recording sites of Cz and Mi, waves II, III, and V are significantly different (p< .01) at the two sites ( Table 2). Waves IIIn and IVn were not evaluated because of insufficient data.
Wave IVn was not defined in three of the nine adult subjects and is excluded from statistical analysis (Table 1). Significant latency differences (p < .02) exist for waves In , II , IIIn, V, and Vn (Table 2) between Mc and Cz recording sites.
The Na recording site was similar in waveform polarity to Mc and Cz (Figs 1 and 2). Except for wave II, the Na recording site tended to have more variability for each of the ABR components than did the other recording sites. Significant latency shifts (p< .02) for waves III and V were noted for Na as compared to Cz ( Differential Recordings. Differential recordings between Cz and Mi from one individual are illustrated in Fig 2. The major components of the ABR, except for wave IVn, are readily identifiable. Wave I is large in the Cz-Mi recording and this may account for its low variability of latency compared to the noncephalic referential recordings.  The ABR components for the Cz-Mc differential recordings had inconsistent waveform identification for waves II , Iln , IV , and Vn. Wave I was observed as being small in amplitude. This is because the components at the two electrode sites (Cz and Mc) are of similar polarity, latency, and amplitude at this time domain. Consequently, by nature of d ifferential recordings, the amplitudes are reduced.

TABLE 1. MEANS AND STANDARD DEVIATIONS (MS) OF ABR COMPONENTS IN NINE ADULT SUBJECTS
The Mc-Mi differential derivation had significant latency shifts (p< .001, Table 2) compared to the Cz-Mi derivation for wave II . There is a considerable decrease in latency for waves V and Vn for Mc-Mi as compared to the other recording conditions (Table 1). In contrast, w ave IIIn has a consistently longer latency for the Mc-Mi derivation than for the other recording sites.     Tables l and 3). Filtering between 150 to 3,000 Hz (Fig 6) enhances the peaks, especially for waves III and V, and the wave IV to V complex becomes narrowed.
In the neonate, waves In, Iln , Illn, and Vn seen at Mi could not be easily identified compared to the adult potentials. Wave I at Mi was somewhat shorter in latency (0.095 ms) than wave I at Cz. Waves II , III, and V demonstrated significant latency differences between Cz and Mi ( Table 4) . The absolute differences in latencies between these two recording sites are greater in the infant than in the  The morphology of the Na recording is very similar to that seen for the Cz recording except for an increase in variability. Waves IV and IVn are generally not identifiable in the neonatal population from the Na recording site, whereas, in the adult population, only IVn could not be consistently identified at this recording site. Latency differences between Cz and Na recording sites were not significantly different. One of the more dramatic differences, other than the overall latency shift between the adult and neonate, is the amplitude and long duration of a positive potential shift (y) occurring between waves II and III defined at Mi (Figs 1 and   utilizing ABR for estimating cochlear function and for defining apparent central conduction times to assess brain stem function. As with the adult,1 wave I in the neonate is positive at Mc and Na, and negative at Mi , and may be described as vectors oriented in both t he horizontal and vertical planes originating close to the ipsilateral ear. Picton et al 5 and others i.J.u .•. 9 · 12 have suggested that wave I at the ipsilateral mastoid is identical to the whole eighth nerve action potential of N l recorded at the round window .
Wave I generally has lower amplitudes for lateral-posterior electrode sites in referential recordings. Consequently, subject variability probably accounts for differences between the individual trac- ... in gs (Fig 2), grand average tracings of wave I (Fig  ms) than in the adult (0. 2 ms). This shift is probably 1), and perhaps wave x in the adult.
related to the absence of the x component at Mi and Hughes et al' identified wave In at Mi as being positive, and negative at Mc and Cz with the largest amplitude at Cz. Starr and Squires 1 were not able to consistently observe In at Mi, but similar to Hughes et al' found wave In to be greatest at Cz. The presence of a positive peak (x) at Mi in our data is similar to Hughes's' observations of a positive peak occurring after wave I at Mi. However, in our results this positive peak, x, is distinct from wave In which is negative at Mi. Hughes et al' also show a negative peak at Mi prior to wave II, which we identify as wave In.
Wave II is positive at all electrode locations in the adult 10 as in the neonate except at Mi where it is generally not observable. The adult tracings of wave II were much more stable for both amplitude and latency, whereas, the neonatal group showed greater variation. Starr and Squires' suggest that in the adult, wave II can be represented as a dipole oriented in the saggital plane. Because of the variability of wave II in the neonate, it is difficult to determine if the same generators are operational in both the adult and neonate.
Wave Iln is seen as a negative deflection at each electrode position in the adult, and in the neonate, the same is true except at Mi and Mc where either an extremely small amplitude for Iln is observed or it is absent altogether. In the adult, waves II and Iln are very similar in scalp distribution and may reflect the same generators. 1.• In the neonate, wave II is more easily identified than wave IIn at Mi. Whether this is due to the features of the developing neonatal brain or to the separation of two different generators for wave II and IIn, cannot be ascertained; however, we favor the former postulate.
The definition of a high amplitude component at Mi occurring between waves Iln and III (labeled y in our recordings) may account for the difficulty, particularly in the neonate, in detecting wave II using the standard Cz-Mi recordings. In neonatal ABRs derived from Cz-Mi there is a large negative shift between waves I and III that obliterates wave II." In contrast, in the adults the negative shift is slight and wave II is clear. The negative shift in the neonate is, in fact, the y positive component defined at Mi which in the differential recording, Cz-Mi, becomes negative. The amplitude of this shift is very large compared to wave II, thereby rendering the detection of wave II difficult.
Wave III occurs consistently as a negative deflection at Mi and a positive deflection at other electrode locations. This observation is in agreement with that of Picton et al5 and Hughes et al, 9 but does not agree with Kevanishvili 8 who found wave III to be absent at Mi. There is more of a latency shift of wave III at Mi compared to Cz in the neonate (0.4 the reduced amplitude of the slow potential in the neonate. It has been suggested that wave III has both a vertical and horizontal dipole component 1 and is quite broad in its spatial extension. Our results agree with these findings. The neonate shows a positive polarity for wave III except at Mi where it is negative and peaks at an increased latency relative to the other recording sites. Wave II is longer in duration for both the adult and neonate in the Mc-Mi derivation supporting the concept that this wave has a spatial distribution as a horizontally oriented dipole. Wave Illn is similar to wave III. Although wave Illn was negative in the neonate, it had a much reduced amplitude at Mc and was generally not observed at Mi. The similarity in the behavior for waves III and Illn may suggest that 1) a common generator(s) produces waves III and IIIn such that their dipoles are broad enough to produce a latency shift due to the spatial distribution of the two poles, or 2) two parallel, but separate, generators exist that are separated in space with similar orientations.
Wave IV had its longest mean latency at Mi and was not statistically different in latency from other electrode locations in either the adult or neonate. This was also observed for wave IVn. This finding is in contrast with that of Starr and Squires' and with the Kevanishvili' observation of increased latency at Mc, thereby suggesting a vertically oriented dipole. The morphology of waves IV and IVn with our wide filter setting are not consistently identifiable.
Wave V was positive at all scalp locations in both the adults and neonates. In addition, significant (p < . 02) differences in latencies occurred between recording sites except for Cz versus Na in the neonate. Furthermore, there is a greater shift in latency differences between electrode sites in the neonate than in the adult. In both groups the shortest latency was observed at Mc. The amplitude of wave Vis clearly largest at Cz, suggesting vertically oriented dipoles. Unlike Starr and Squires, 1 we did not observe an absence of wave V for the Mc-Mi derivation, but we did observe the latency disparities between the two sides of the brain stem (Mc versus Mi) at a significant level of difference (p< .001).
Wave Vn was negative at all scalp locations in both groups with the shortest latency at Mc. Wave Vn behaved quite similarly to wave V and may have similar origins.
It has been proposed by Robinson and Rudge' 0 that, when there is a shift in the latency of a component of the ABR, one of three possibilities exists: 1) there is more than one active generator within a given pathway, 2) there is more than one pathway being activated in the generation of the waveform, or 3) that both of the above cases occur. Robinson and Rudge 10 confirmed earlier observations of Pie-ton et al5 relative to differences between vertex-ipsi-contributing to the many components of the ABR. lateral mastoid recordings (Mc-Mi} and vertex-con- The observations of statistically significant differtralateral mastoid recordings (Cz-Mc). It was their ences of latency across the head (Mi-Cvu and Mccontention that this could be accounted for by the Cvn) in the two populations (adults versus neopresence of multiple generators, and they presented nates} could be due to l } the effects of maturation evidence for the existence of at least three sources on the neural generators of the ABR, 2} the exiscontributing to the ABR. In addition to the above tence of separate generators in the neonates and the considerations, it may be that the latency shift is ac-adults, or 3} a combination of both. As the nervous tually a partial phase shift in a dipole generating the system matures, myelinization occurs with a comcomponent due to structural changes accompany-parable decrease in both the absolute latencies of ing maturation, resulting in a reorientation of the the ABR components and the scalp differences. As dipole. Certainly, as the head size and shape change part of the maturing process the weighting of the from infant to adult, boundary conditions for the neural components contributing to the various current fields recorded at the scalp change. Since peaks may alter. Secondly, as the physical size of wave I is essentially the same in both the neonate the brain and surrounding structures increase in and adult (except for a latency shift}, it is unlikely, size, the orientation of the generators might in our opinion, that differences in cochlear organi-change, with alterations in their relative polarities, zation or timing would influence later waveforms eg, orientation of the dipoles, relative to the scalp. without affecting wave I.
One would expect then, by carefully following The results of our present investigation would likewise suggest the existence of multiple generators growing infants, to see gradual changes of the latency differences between various recording montages toward the adult type ABR.