www.elsevier.com/locate/ynimg NeuroImage 33 (2006) 759 – 773
Frontal lobe involvement in the processing of meaningful auditory stimuli develops during childhood and adolescence Stephan Bender, a,b,c,⁎ Rieke Oelkers-Ax, a Franz Resch, a and Matthias Weisbrod b,d a
Department for Child and Adolescent Psychiatry, Psychiatric Hospital, University of Heidelberg, Blumenstraβe 8, D-69115 Heidelberg, Germany Section for Experimental Psychopathology, Psychiatric Hospital, University of Heidelberg, Voβstraβe 4, D-69115 Heidelberg, Germany c Psychosomatic Hospital, University of Heidelberg, Im Neuenheimer Feld 410, D-69120 Heidelberg, Germany d SRH-Klinikum Karlsbad-Langensteinbach, Department of Psychiatry and Psychotheraphy, Guttmannstraβe1, D-76307 Karlsbad, Germany b
Received 30 January 2006; revised 18 May 2006; accepted 3 July 2006 Available online 23 August 2006 Auditory event-related N1b reflects attention-related processing in bilateral temporal auditory cortex. Frontal contributions indicating an orienting reaction have been suggested. We analyzed the maturation of frontal contributions to the auditory event-related potential following the warning stimulus in a contingent negative variation (CNV) task by high-resolution current source density mapping and spatio-temporal source analysis in 80 healthy subjects and 121 primary headache patients (migraine with/without aura, tension type headache) from 6 to 18 years; as increased orienting responses and disturbed maturation have been described in migraineurs. A selective local increase of N1b with age occurred at midfrontocentral leads. This increase could not be explained sufficiently by overlapping bilateral temporal sources but pointed towards additional frontal activation over the supplementary motor area (SMA) in adolescents which was absent in children. A second frontal N1 component peaked about 50 ms later, showed an earlier maturation and has been suggested to reflect early response selection processes in the anterior cingulate. Primary headache patients showed the same component structure and developmental trajectory as healthy subjects without significant influences of differential diagnosis. We conclude that: (1) Brain maturation crucially influences N1b. (2) Two frontal lobe N1 components can be dissociated in their maturational trajectory. (3) Early SMA activation could be elicited by rare auditory stimuli from about 12 years on, allowing fast sensorymotor coupling without previous categorical stimulus classification. (4) Primary headache patients did not differ in their maturation of frontal or temporal contributions to N1b when elicited by moderately loud short tone bursts. © 2006 Elsevier Inc. All rights reserved. Keywords: Event-related potential; N1; Auditory; Age factors; Current source density; Source analysis; Frontal lobe; Primary headache; Migraine
⁎ Corresponding author. Department for Child and Adolescent Psychiatry, University of Heidelberg, Blumenstrasse 8, D-69115 Heidelberg, Germany. Fax: +49 6221 56 69 41. E-mail address: [email protected]
(S. Bender). Available online on ScienceDirect (www.sciencedirect.com). 1053-8119/$ - see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2006.07.003
Introduction Auditory reaction time decreases with increasing age until adolescence with a much steeper slope than central motor conduction time (Fietzek et al., 2000). Most variability of reaction time is ascribed to the efficiency of the coupling between sensory perception and motor response execution (Menon et al., 1998). In the current study we wanted to investigate whether the decrease in auditory reaction time during childhood and adolescence could be in part due to the maturation of a more effective “short-cut” from auditory areas to secondary motor areas involved in movement programming. The auditory-evoked potential is able to image cortical activation to behavior-relevant auditory stimuli in the millisecond range. Its N1 component consists of at least three different subcomponents (N1a, N1b, N1c) (Naatanen and Picton, 1987) and is an electrophysiological indicator for auditory perception and attention. Both, N1a and N1c, show a bilateral temporal maximum on the scalp surface and most likely originate from radial activity in the lateral auditory association cortices (Giard et al., 1994). In contrast, N1b peaks over central areas around the vertex (e.g., Alcaini et al., 1994). N1b in adults is thought to involve bilateral major sources in the temporal auditory cortex (tangentially oriented dipoles in the supratemporal plane, Heschl gyrus, superior temporal lobe, which project to the central scalp surface) as well as a smaller radial contribution from the frontal lobe (premotor cortices, supplementary motor area (SMA) or anterior cingulate (ACC); Alcaini et al., 1994; Giard et al., 1994; Naatanen and Picton, 1987; Picton et al., 1995). The involvement of the frontal lobe is thought to represent an orienting reaction while temporal sources are supposed to reflect auditory attention-related stimulus processing (Naatanen and Picton, 1987). Auditory N1 develops with age during childhood and adolescence: N1a and N1c have been found to decrease in amplitude with increasing age while N1b becomes larger with age in most studies (Ceponiene et al., 2002; Gomes et al., 2001; Goodin et al., 1978; Pang and Taylor, 2000; Ponton et al., 2000; Tonnquist-Uhlen et al.,
S. Bender et al. / NeuroImage 33 (2006) 759–773
2003). Furthermore, children show a central N1b only with interstimulus intervals of a couple of seconds or longer (Ceponiene et al., 2002). The question about the origin of the N1b increase during childhood and adolescence and especially the role of frontal and temporal contributions to this increase has not been resolved yet. Frontal contributions to age-dependent topography changes of N1b must be distinguished from changes in temporal auditory cortex activity (Gomes et al., 2001). Frontal lobe maturation is delayed, both morphologically (Huttenlocher, 1990) as well as functionally (Andersen, 2001; Fischer et al., 1997; Luciana and Nelson, 1998). Auditory attention includes several aspects with different maturational trajectories (Gomes et al., 2000). N1b with its supposed frontal (orienting, motor) and temporal (sensory) contributions is an excellent tool in order to investigate the influences of delayed frontal lobe maturation on auditory perception, orienting, fast response selection and frontal attentional fine-tuning of posterior sensory responses. The understanding of an age-dependent frontal contribution to auditory N1 seems important, since it contributes to understand the developmental maturation of the functionally relevant interaction between frontal and sensory areas in the analysis of rare meaningful stimuli. Why do adults react substantially faster than children in simple reaction time tasks? Is there a “short-cut” in the mature brain which allows fast reactions without previous stimulus classification when we note that we have heard something (due to late frontal lobe maturation or myelination; e.g., Steen et al., 1997)? A modality unspecific orienting reaction could also modulate sensory cortex activity via frontal areas (Amassian et al., 1998). We hypothesized that a still missing sensory/motor short-cut in children would be reflected by lower or missing frontal contributions to immature auditory N1. Alternatively, developmental changes in N1 would have to be ascribed to auditory cortex maturation. Moreover, a second frontal component has been described in adults peaking around 40 ms after N1b (Mulert et al., 2001; Alcaini et al., 1994). It has been attributed to anterior cingulate activation reflecting early response selection processes and has been found diminished in schizophrenic patients (Mulert et al., 2001). The maturation of this second frontal component during childhood and adolescence has not been investigated so far. Overlays between this second N1 component with N1b must be taken into consideration when describing age-related N1 changes and, more important, the maturational trajectories of different frontal N1 contributions (reflecting most likely an orienting reaction and early response selection) have to be compared. In the stimulus–intensity slope (ASF slope) of adult auditory N1 (Wang et al., 1996), differences between migraineurs and healthy subjects have been found. These differences have been related to lower central serotonergic neurotransmission in the temporal auditory cortex in migraineurs (Hegerl et al., 1994; Wang et al., 1996). However, a deficient habituation in migraineurs might contribute substantially to alterations in ASF slope (Ambrosini et al., 2003). Thus also the relative contributions of frontal (orienting response) and temporal (auditory cortex) components to auditory N1 amplitude could be altered in migraineurs, accounting for differences in ASF slope. Correlates of the orienting reaction and movement-related potentials and such as early and late CNV (Bender et al., 2002; Gerber and Schoenen, 1998; Kropp et al., 1999) and the Bereitschaftspotential (Muller et al., 2002) have been found to be increased in migraine children and adult migraineurs. Moreover, there are hints towards a disturbed cerebral maturation in migraine children (Bender et al., 2002; Kropp et al., 1999; Muller et al., 2002; Oelkers-Ax et al., 2004). We thus examined whether
there would be further hints towards disturbed maturation in children with primary headache, e.g., due to larger frontal contributions to N1 at earlier ages, or if these children and adolescents would confirm the same maturational pattern as healthy subjects in a second independent sample. Methods Subjects 81 healthy children and adolescents (42 male and 39 female, mean age 11.6 ± 3.3 years) and 123 children and adolescents with primary headache (according to the criteria of the International Headache Society IHS 1988), 67 migraineurs without aura, 11.0 ±3.0 years, 40 male and 27 female, 32 migraineurs with aura, 12.8 ±3.4 years, 14 male and 18 female, 24 tension type headache patients, 12.0 ± 3.1 years, 12 male and 12 female) between 6 and 18 years were recruited for CNV recordings in a study concerning headache in childhood and adolescence. First results concerning effects of migraine diagnosis on CNV maturation at electrode Cz have been published elsewhere (Bender et al., 2002). EEG recordings for CNV could be completed in 80 healthy and 121 headache subjects. Subjects were right-handed except for 6 healthy children and 12 children with primary headache (assessed by the Edinburgh Handedness Inventory; Oldfield, 1971). All headache subjects were pain-free for at least 72 h before the recording (Siniatchkin et al., 2000). 22 children of the headache group reported pain within 72 h after data recording. For 5 children the recording session could be repeated and they were pain free afterwards. We compared the presented analyses with and without the exclusion of the 17 subjects for which the session could not be repeated and found no significant changes (not shown). Also the left-handed subjects did not critically influence the presented results and were not excluded. No subjects had to be excluded because of artifacts within the auditory ERP time range. Subjects were screened for clinical hearing impairments, neurological (other than headache) or psychiatric diseases. None took any psychoactive drugs. The study was approved by the local ethics committee and all participants and their parents provided informed written consent according to the Declaration of Helsinki. Recording/Data pre-processing We recorded 60 trials in three consecutive blocks of 20 CNV trials using a short (Vidal et al., 2005) diotic tone burst as warning stimulus S1 (frequency, 1000 Hz; duration, 50 ms; 90 dB, near instantaneous rise time) and a short diotic tone burst as imperative stimulus S2 (frequency, 2000 Hz; duration, 50 ms; 90 dB, near instantaneous rise time; Neuroscan Stim Gentask; Neuroscan Inc., TX, USA). Short tone bursts allow a good assessment of auditory alerting by behavior-relevant stimuli. We have analyzed the auditory event-related potential following the warning stimulus S1. Intertrial intervals (the interval before S1) varied randomly from 10 to 15 s, stimulus onset asynchrony between S1 and S2 was 3 s. Subjects were instructed to respond as fast as possible when S2 occurred by pressing a mouse button with the index finger of the dominant hand. Participants fixated a cross on a computer screen at 1-m distance. Neuroscan Synamp Amplifiers (Neuroscan Inc., TX, USA) were used to record continuous DC 64-channel EEG with a sampling rate of 250 Hz. Electrodes were fixed using an equidistant electrode cap (Easycap, FMS, Germany) and are
S. Bender et al. / NeuroImage 33 (2006) 759–773
named according to the equivalent positions in an extended international 10–20 system (small deviations are indicated by ’). Vertical and horizontal electrooculogram were also recorded from electrodes next to the outer canthi and 1 cm above and below the left eye. Impedances were kept below 5 kΩ. Data were recorded against a reference near the vertex and transformed offline to average reference. Recordings 1 s before S1 served as baseline. The EEG signal was digitally filtered (30 Hz high cut-off), segmented into epochs of 7.5 s (1 s pre S1 to 3.5 s post S2), corrected automatically for DC drifts by linear regression analysis (BrainVision Analyzer, BrainProducts GmbH, Munich, Germany), for eye movements and blinks (algorithm described by Gratton et al., 1983). Artifacts were rejected automatically if the signal amplitude exceeded 150 μV. This procedure was confirmed by visual inspection, remaining artifacts were removed manually. Less than one third of trials was removed. Bad channels (less than 1%) were interpolated using nearest neighbors. Analysis Amplitudes, latencies N1b amplitude for average referenced data and CSD was determined as the mean value in a 24-ms time window around the maximum at Cz between 80 and 160 ms post S1 with respect to baseline. N1b latency was calculated as the time between onset of S1 and the N1b peak at Cz (Ceponiene et al., 2002; Pang and Taylor, 2000). The N1c peak was determined as the local maximum at T7 between 100 and 250 ms post S1 (Giard et al., 1994; Gomes et al., 2001; Tonnquist-Uhlen et al., 2003) and peak latency was calculated as the time between S1 onset and the N1c peak. N1c amplitude was determined as the time window 48 ms around the N1c peak (Ceponiene et al., 2002; Gomes et al., 2001; Naatanen and Picton, 1987; Pang and Taylor, 2000); N1c is a broader peak than N1b. P2 latency was determined as the positive peak at Cz between 160 and 260 ms (Sheehan et al., 2005) in order to determine possible overlaps with N1c. Latencies and amplitudes of average-referenced data and CSD values of N1b and N1c were tested for continuous maturation effects by linear regression analysis. For most parameters, the linear model was a sufficient approximation though it does not reflect all characteristics of physiological maturation (often decelerated asymptotic trajectories) (Bender et al., 2005). After describing continuous maturation, the sample was split in half into 6–11- and 12–18-year-old subjects for the following analyses in order to achieve a good signal-to-noise ratio and to test for interactions between age and diagnosis-related effects (Oelkers-Ax et al., 2004). An ANOVA with the factors diagnosis (healthy, migraine without aura, migraine with aura, tension type headache) and age (6–11 and 12–18 years) was performed in order to test for diagnosis-related differences at Cz. The current sink at Cz was tested for significance against pre-stimulus baseline during the N1b time window in healthy subjects and headache patients for 6–11and 12–18-year-old subjects by t-tests. Topographic analysis Qualitative changes in N1b topography during maturation in childhood and adolescence offered the unique opportunity to dissociate different components of N1b. Topographic analysis was carried out in several steps by application of complementary methods in order to elucidate different aspects (Bender et al., 2005):
In a first step we analyzed average referenced data. Average referenced data reflect both local activity as well as volumeconducted activity from distant sources and minimize effects related to the reference electrodes. An ANOVA with the independent factors age group (6–11 and 12–18 years), diagnosis (healthy subjects versus primary episodic headache) and electrode location (Cz, C3, C4; repeated measurements) was used in order to obtain first hints whether a qualitative change in N1b occurred during development. For post hoc analyses Scheffe’s tests were used. In the case of a selective developmental midfrontocentral negativity increase (additional local frontal activity) we expected an interaction between the factors age group and electrode location with Cz showing a more pronounced increase in negativity than C3 and C4. While in average referenced data a midline central maximum can arise from summation of bilateral sources, scalp current density can potentially resolve this problem: in a second step we calculated current source density (CSD) in order to describe the development of local frontal cortical activity during N1b with the highest possible spatial resolution attenuating effects of volume conduction. CSD emphasizes local cortical activity over widespread volume conducted activity by virtually referencing each electrode against the surrounding potential distribution and thus acting as a spatial filter. In the case of exclusive temporal auditory cortex activity we expected two independent (separated in the midline) bilateral combinations of a central current sink and a temporal current source reflecting the tangentially oriented dipoles in the auditory cortex (Nunez et al., 1991). In the case of frontal activity we expected an additional midline frontocentral current sink (Giard et al., 1994). In a final step we used spatio-temporal source analysis (BESA, Megis GmbH, Gräfeling, Germany) in order to test whether frontal radial or bilateral temporal tangential sources would better explain age-related midcentral N1b increases. Dipole analysis is the method of choice to integrate information from distant leads on the skull surface in one model and takes advantage of volume conduction. In the case of increased tangential temporal activity, a corresponding deep temporal positivity should develop in a parallel way to central N1b increases, resulting in increased dipole moments of tangential temporal sources. By source analysis, we could also include the time dimension into analysis in order to dissociate two possible frontal contributions (N1b and a second frontal peak about 40 ms later) not only in topography but also in their time course to exclude effects of temporal overlays between both components. CSD analysis in the N1b time window. CSD maps using a spherical spline interpolation algorithm (Perrin et al., 1989) (smoothing constant 10− 6) were calculated at the central N1b peak. Continuous maturation was described by current source density mapping for 6–7-, 8–9-, 10–11-, 12–13-, 14–16- and 17– 18-year-old subjects according to published ERP analysis guidelines (Picton et al., 2000). Overall developments were analyzed for pre- and (post)pubertal subjects (6–11 and 12–18 years) (Bender et al., 2005; Oelkers-Ax et al., 2004). CSD analysis of the second frontal component. A second frontal N1 component peaking about 40 ms after N1b has been supposed to reflect early anterior cingulate activity (Alcaini et al., 1994; Mulert et al., 2001). There was no a priori hypothesis available where it would peak exactly in children and adolescents because it has been reported so far only in adult subjects. So in order to test whether this frontal component would be present during childhood and adolescence and in order to describe its topography, CSD
S. Bender et al. / NeuroImage 33 (2006) 759–773
values at the peak at Fz in the time window reported in literature (CSD amplitude 148–180 ms post S1, about 40 ms after central N1b (Alcaini et al., 1994; Mulert et al., 2001), which was in good agreement with the latency of the CSD-peak at Fz of about 165 ms in our data) were tested for significance using Bonferroni correction (p < 0.00078) in all electrodes. Spatio-temporal source analysis (BESA). In a final step, we applied a source analysis approach using multiple spatially fixed temporally variable dipoles in order to separate both frontal N1 components in their time course. The dipole model was established as follows (Picton et al., 1995): the dipole model was fitted first on the grand average of the age group 12–18 years, because grand averages show a better signal-to-noise ratio than single subject averages and allow more stable dipole solutions. Two symmetrical dipoles were fitted on the ascending part of N1b (in order to model the vertically oriented activity in bilateral temporal auditory cortices) because there was the least overlap between N1b and N1c in this time window. Second, two symmetrical dipoles were fitted on the maximum of N1c (in order to model radially oriented activity in bilateral temporal auditory cortices). It has been described in literature, that this way tangential (step 1, N1b) and radial (step 2, N1c) components of mature N1 arising in temporal auditory cortex can be well modeled and separated (Picton et al., 1995; Scherg and Von Cramon, 1985; Vaughan et al., 1980). In order to test if there would be additional activity, a regional source was fitted on the whole N1 interval. A regional source is a combination of three single dipoles oriented in all three axes of 3dimensional space. So regional sources are especially suitable for scanning for unexplained activity of any orientation of a certain brain area. It is important to state that the localization of the regional source has no implications about the exact localization of the generators of the explained activity. The interval 108–132 ms was applied in order to calculate the residual variance of the model. By converting the regional source to single dipoles (with a single spatial orientation) and fitting the single dipoles on the intervals of the respective two main peaks of the regional source waveforms, we determined which waveform of the regional source belonged to which spatial orientation. This was possible, because the source waveforms for the three spatial orientations of the regional source showed little temporal overlap and the two frontal N1 components could be separated this way (see dipoles #5 and #6 in Fig. 5—same location but different orientation): the light green dipole #5 was fitted on an interval during the N1b maximum at the vertex; the dark green dipole #6 was fitted on an interval when the second frontal N1 peak reached its maximum about 40–50 ms later (according to the two main peaks in the regional source waveforms). There was no considerable activity in the third axis of the regional source. The introduction of the additional frontal source did not crucially influence the source waveform activity of the temporal dipoles but explained additional signal variance. The established model was later applied to the grand average of 6–11-year-old subjects in order to compare the age-dependent development of the respective source activities. For the comparison of the contribution of frontal activity, the temporal dipoles were refitted on the grand average of the age group 6–11 years (same procedure as described above for 12–18-year olds) so that differences in the amount of activity taken up by the frontal regional source could be distinguished from developmental changes in location and orientation of the temporal dipoles. We made sure, that refitting the temporal dipoles did not produce other
alterations in the model but only an optimization resulting in small changes in location and orientation. As there was no additional variance during N1b which could be explained by a frontal regional source in 6–11-year-old children, its location was kept constant. The same N1b interval was applied for the calculation of the residual variance of the model for 6–11-year olds as N1b latency was only a few milliseconds longer for these younger children (see below). In order to test for significant frontal contributions to N1b, the models for 6–11- and 12–18-year-old subjects (optimized location of the temporal dipoles for each age group while the location of the frontal sources was kept constant) were applied to individual subjects after refitting each dipole’s orientation (same fit intervals as for the model on the grand average) on the individual averages. We made sure that applying the same model with constant locations of the temporal sources to all subjects yielded almost identical results with respect to the development of the frontal N1b contribution so that slightly different locations of the temporal dipoles in the models for 6–11- and 12–18-year-old subjects did not produce an artificial development in the frontal dipole’s activity. The frontal light green dipole’s (#5 in Fig. 5, oriented towards the vertex) dipole moment in the window 24 ms around the individual N1b peak at Cz was exported and a possible agerelated development was tested by linear regression analysis. We tested for significant contributions of the frontal source to N1b by t-tests against baseline for 6–11- and 12–18-year-old subjects (healthy subjects as well as headache patients). Results N1b Average referenced data We replicated findings about overall N1 development: we found a slight decrease in latency with age for N1b at Cz: linear regression equation 128 ± 4.3 ms (± standard error) − 0.72 ± 0.356 ms/year; t = 2.0; p = 0.045 for healthy subjects and 126 ± 3.1 – 0.51 ± 0.26 ms/year; t = 2.0; p = 0.05 for headache patients). Mean N1b amplitude at Cz increased significantly with age (see Tables 1 and 2). Table 1 N1b and N1c amplitude development (± standard deviation) against average reference Age (N)
Amplitude [μV] N1b (Cz)
Amplitude [μV] N1b (C3/C4)
Amplitude [μV] N1c (T7)
Healthy subjects 6–7 (14) − 1.6 ± 3.6 8–9 (16) − 6.6 ± 9.2 10–11 (9) − 7.1 ± 1.9 12–13 (22) − 8.0 ± 5.1 14–16 (13) − 9.0 ± 5.6 17–18 (6) − 12.4 ± 6.1
− 5.1 ± 2.9 − 5.7 ± 5.4 − 5.9 ± 1.8 − 6.8 ± 2.8 − 6.9 ± 4.1 − 8.8 ± 3.1
−11.0 ± 4.6 − 10.4 ± 4.4 − 9.0 ± 5.2 − 8.0 ± 3.7 − 6.7 ± 3.9 − 4.4 ± 2.3
Headache patients 6–7 (13) − 2.4 ± 5.1 8–9 (28) − 6.3 ± 4.1 10–11 (30) − 6.1 ± 4.4 12–13 (21) − 8.3 ± 5.9 14–16 (20) − 9.8 ± 4.9 17–18 (9) − 8.8 ± 5.4
− 3.8 ± 4.0 − 6.4 ± 2.9 − 6.5 ± 2.9 − 7.0 ± 3.5 − 6.6 ± 2.6 − 5.0 ± 2.3
− 10.3 ± 6.9 − 9.0 ± 4.6 − 7.1 ± 3.6 − 7.5 ± 3.2 − 4.4 ± 3.3 − 3.4 ± 2.0
S. Bender et al. / NeuroImage 33 (2006) 759–773
Table 2 Regression slopes (±standard error) when N1b amplitude is predicted by age Healthy subjects
Cz C3 C4
− 0.80 ± 0.2 μV/year − 0.17 ± 0.12 μV/year − 0.35 ± 0.15 μV/year
t = 4.0 (p = 0.0001) t = 1.4 (p = 0.17) t = 2.3 (p = 0.02)
−0.59 ± 0.14 μV/year −0.01 ± 0.1 μV/year −0.20 ± 0.11 μV/year
t = 4.2 (p < 0.0001) t = 0.2 (p = 0.88) t = 1.8 (p = 0.07)
The grand averages for 6–11- and 12–18-year-old healthy subjects against an average reference are shown in Fig. 1 (headache patients exhibited a very similar topography). Please note the decrease of N1c at temporal leads and the increase of N1b which appears more pronounced at Cz than at C3 or C4. Continuous age-dependent development is presented in Table 1 (small age groups, separated for healthy subjects and headache patients, cf. also Fig. 2) as well as in Fig. 3: Fig. 3 displays a scatterplot of the development of N1b amplitude for electrodes Cz
and pooled leads C3/C4 for both healthy and headache subjects, demonstrating that the more pronounced increase of N1b amplitude at Cz than at C3/C4 was the result of continuous maturation with considerable inter-individual variability. The most pronounced qualitative changes occurred between 6–7- and 8–9year-old children (see Table 1 and Fig. 3). For the statistical significance of age-dependent development of average referenced amplitudes at central leads (Cz, C3, C4) see Table 2 (linear regression slopes).
Fig. 1. Topography of the auditory event-related potential after S1 (auditory warning stimulus)—average referenced data of healthy subjects: 6–11 years (red line), 12–18 years (black line), negative is up. Headache patients showed a similar topography. Please note the marked increase in N1b over the vertex, the slight N1b increase around C3, C4, and the bilateral decrease of temporal N1a and especially N1c with increasing age. Please also note that the N1 peak at Fz is lasting longer than at Cz. EOG is given at a different scaling. The vertical dashed line indicates the onset of the tone burst (S1), 100 ms pre-stimulus served as baseline. The interval 100 ms before S1 to 300 ms after S1 is shown. Please note that in this figure electrodes are arranged like concentric circles around Cz and that the lowest electrodes (F9, F10, FT9′, FT10′, TP9′, TP10′, P9′, P10′, O9′, O10′, Iz) were located below the equator (polar coordinates azimuth θ = 115/− 115).
S. Bender et al. / NeuroImage 33 (2006) 759–773
Fig. 2. Upper two rows / Lower two rows: isopotential line CSD maps during the N1b peak for the age groups 6–7, 8–9, 10–11, 12–13, 14–16 and 17–18 years for healthy subjects (top) and headache patients (bottom). Dotted blue areas indicate current sinks (negativity), red isopotential lines indicate current sources (positivity), scaling: 0.7 μV/cm2 per isopotential line. Please note the parallel development for healthy controls and headache patients resulting in an increase in midfrontocentral negativity (dotted blue area) in both groups while the bilateral centro/temporal tangential dipole pattern consisting of a central current sink and a temporal current source does not change much. The tangential dipole pattern on the right side seems to be stronger developed in both healthy and headache subjects. Isopotential line CSD maps for 6–11 (left)- and 12–18-year-old subjects (right). Top: healthy subjects, bottom: headache patients. Please note how the increased signal-to-noise ratio (bigger number of subjects who form the averages) emphasizes the characteristic changes during maturation of N1b. Same scaling.
In the ANOVA with the factors headache (headache versus healthy subjects), age (6–11 and 12–18 years) and electrode (Cz, C3, C4; repeated measurement) we obtained highly significant main effects for age (F = 12.8; p = 0.0004) and electrode (F = 8.1; p = 0.0003) as well as a highly significant interaction between age and electrode (F = 17.9; p < 0.0001) indicating a different agedependent development of N1b at Cz/C3/C4. In contrast, dipole simulation (Dipole Simulator, BESA, Megis GmbH, Germany) yielded a similar negativity increase at Cz/C3/C4 as well as an increased low temporal positivity when dipole strength of bilateral distant tangential temporal sources was augmented (not presented in detail, see results of source analysis below). Scheffe’s post hoc test was conducted and revealed significant differences between N1b amplitude at Cz (mean −9.0 μV) and C3 (mean −6.0 μV; p = 0.004) as well as Cz and C4 (mean −7.3 μV; p < 0.0001) for 12–18 years but not 6–11 years (Cz −5.5 μV versus C3 −5.8 μV; p = 0.89 and Cz versus C4 −6.0 μV; p = 0.98) indicating that no increase comparable to the one at Cz was found at C3/C4. While for 6–11-year-old children all central electrodes showed similar N1b amplitudes, for 12–18-year-old adolescents the N1b amplitude at Cz highly significantly exceeded the amplitudes at C3 and C4. All reported differences were also significant when the healthy subjects were analyzed without taking into account the results of headache patients. Healthy subjects and subjects suffering from headache showed a parallel age-dependent development. There was no effect of headache on N1b amplitude—main effect headache, F = 0.005;
p = 0.94. Neither was there a significant effect of headache differential diagnosis on N1b amplitude at Cz (ANOVA with factor differential diagnosis – migraine without aura, migraine with aura, tension type headache, healthy controls – age group and electrode): main effect differential diagnosis, F = 1.4, p = 0.26. Current source density analysis: maturation of the midfrontocentral current sink Current source density analysis was carried out in order to isolate local cortical activity from widespread potentials produced by volume conduction from distant areas. Fig. 2 illustrates the development of CSD maps (averages of individually calculated CSD) of healthy subjects and headache patients for 6–7-, 8–9-, 10– 11-, 12–13-, 14–16- and 17–18-year-old subjects in order to show continuous maturation and for 6–11- and 12–18-year-old subjects in order to show the most reliable maturational changes with increased signal-to-noise ratio (39 healthy subjects 6–11 years and 41 healthy subjects 12–18 years; 71 headache patients 6–11 years and 50 headache patients 12–18 years). The CSD maps for subjects from about 12 years on show that the topography of central N1b is qualitatively different from the CSD maps at earlier ages: the bilateral central current sinks around C3/C4 fuse in the midline for older subjects for both healthy controls and headache patients. In contrast to 6–11-year-old children, the current sink at Cz (± standard deviation) was highly significant in the age group 12– 18 years: − 65.1 ± 66.2 μV/m2; t = 6.3; p < 0.00001 for healthy
S. Bender et al. / NeuroImage 33 (2006) 759–773
Fig. 3. Age-dependent development of N1b for both healthy subjects and headache patients—scatterplots and mean values. Top: N1b amplitude against average reference at Cz (one extreme value of a healthy subject, 9 years, 36 μV, is not visible). All diagnosis groups except the tension type headache group (nonsignificant differences, small group) showed a parallel development. Middle: N1b amplitude against average reference at C3 and C4; (C3 + C4)/2. For a separate analysis of C3 and C4 development see Table 2. Bottom: mean values at Cz and (C3/C4)/2 for 6–11- and 12–18-year-old subjects illustrating the significant interaction between electrode location and age group.
subjects and − 58.8 ± 63.2 μV/m2; t = 6.6; p < 0.00001 for headache patients. No topographically clearly separated and independent third current sink appeared reliably in the midline (such as in Fig. 2 for 17–18-year-old headache subjects). Such a third current sink depended highly on interpolation and spatial resolution of CSD mapping (parameters such as the smoothing constant) and was vulnerable to noise.
However, CSD mapping confirmed a continuous qualitative change in N1b topography, which rather selectively affected the midcentral area and left the temporo-central bilateral tangential dipole components (combinations of a temporal current source and a central current sink with a steep potential gradient in between) without systematic changes. The additional midline current sink comprised FCz’ and its surrounding
S. Bender et al. / NeuroImage 33 (2006) 759–773
Fig. 4. Time course of current source density (CSD) during N1b and the 40–50 ms later peak at Cz and Fz as well as CSD topography during N1c (healthy subjects). Top: CSD maps at the peak of N1c, left: 6–11-year-old healthy subjects, right: 12–18-year-old subjects. Scaling: 0.7 μV/cm2 per isopotential line. Blue dotted areas indicate current sinks, red isopotential lines current sources. Please note the similar frontal current sink and the bilateral temporal current sinks for both age groups. Please note that there is a certain overlap with P2. Headache patients showed the same topography. Bottom: CSD time course at Cz (black line) and Fz (red line). Please note, that while midcentral N1b extends into frontal areas in adolescent subjects, there's no hint towards a production of midcentral N1b by an overlap of bitemporal activity with the rising second frontal component which peaks about 40–50 ms later.
electrodes (see Fig. 2)—overlying the supplementary motor area (SMA). N1c N1c amplitude significantly decreased with increasing age (see Table 1) while latency (mean latency 177 ± 27 ms for healthy subjects and 173 ± 28 ms for headache patients) slightly increased: regression equation 150 ms + 2.4 ms/year for healthy subjects (t = 2.7; p = 0.009) and 161 ms + 1.1 ms/year for headache patients (t = 1.3; p = 0.19). P2 latency of healthy subjects was 199 ± 21 ms (headache patients 197 ± 16 ms) and did not change with age (linear regression 198 ms + 0.03 ms/year; t = 0.04; p = 0.97 for
healthy subjects; 200 ms–0.3 ms/year; t = 0.6; p = 0.55 for headache patients). P2 latency was significantly longer than N1c latency (healthy subjects 22 ± 30 ms; t = 6.5; p < 0.00001; headache patients 24 ± 31 ms; t = 8.4; p < 0.00001). Both peaks could be distinguished despite a certain overlap (see Figs. 4 and 5). N1c rather coincided with the second frontal N1 peak than with P2 (see Figs. 4 and 5). Second frontal N1 peak A second frontal current sink could be identified in agreement with literature about 40–50 ms after N1b (see Fig. 4 for the time course of CSD at Fz) with a mean latency of 169 ± 15 ms (SD) for healthy subjects and 168 ± 14 ms for headache subjects. Its latency
Fig. 5. Source analysis of N1 of the auditory even-related potential. Top left: source model for 6–11-year-old healthy subjects. Dipoles 1, 2, 3 and 4 were fitted on the respective N1 peaks (see text) of the grand average of 6–11-year olds. No additional frontal activity was detectable during N1b in 6–11-year-old children. Thus the location of the frontal regional source was kept constant with respect to the adolescents’ model (see below and text) in order to show the development of the amount of activity taken up by the frontal regional source. In order to illustrate the orientation of its two main vectors in 12–18-year-old adolescents as well as the development of their dipole moments in 6–11-year-old children, the regional source was converted to two single dipoles fitted on the respective regional source peaks during the interval of the N1b maximum at the vertex (light green dipole #5) and during the interval when the second frontal N1 peak reaches its maximum about 40–50 ms later (dark green dipole #6). Top right: source waveforms indicating the time course of source activity when the model (top left) is applied to the grand average of 6–11-year-old healthy subjects. Please note that the peak of the frontal source which corresponds to N1b in 12–18-year-old-subjects (light green dipole #5) is missing in 6–11-year olds (indicated by the arrow) while the second frontal peak about 50 ms later (dark green dipole #6) is similar in both age groups. The interval 100 to 132 ms after the onset of the tone burst (warning stimulus, 0 ms) is marked in green in order to compare the different dipoles’ contributions to N1b. Mid left: source model for 12–18-year-old healthy subjects. Mid right: source waveforms indicating the time course of source activity when the model (mid left) is applied to the grand average of 12–18-year-old healthy subjects. Please note the peak in the light green dipole #5 pointing towards the vertex which paralleled N1b. A second peak (dark green dipole #6, pointing towards the midfrontal area) paralleled the second frontal peak about 40–50 ms later. Bottom: scatterplot illustrating the development of the dipole moment [μV] of dipole #5 (frontal contribution to N1b) during the 24 ms around the individual N1b peak at Cz. Please note that only children older than about 10–12 years showed a frontal N1b contribution. One extreme value (9-year-old healthy subject, − 326 μV) is not visible (cf. Fig. 3).
S. Bender et al. / NeuroImage 33 (2006) 759–773
decreased slightly with increasing age over all subjects in a similar manner (regression slope − 1.3 ± 0.3 ms/year). At all ages, the two frontal N1 peaks remained clearly separated. No significant linear age-dependent development was found for the CSD amplitude of this peak at the site of its maximum, Fz, within the examined age range. The maximum amplitude was reached around 10–12 years; if there was any development at all, it followed an inverted U shape with significant activity developing already between 6 and 8 years. Electrodes around the components’ maximum at Fz with significant current sinks at the time of its peak (multiple t-tests with Bonferroni
correction; p < 0.00078) are given in Table 3. Other significant current sinks because of the temporal overlap with N1c (bitemporal) occurred at leads CP4′, C6′, F6′, FT8′, T8, F10, FT10′ and CP5′, C5′, F5′, TP7′, T7, FT7′, FT9′, F9 (p < 0.00078, see Fig. 4). This second frontal component’s topography clearly differed from the fronto-central midline N1b component. There were no hints that the increase in the midfrontocentral current sink of N1b was an artifact resulting from overlap of three different sources (bilateral temporal and frontal, peaking later at Fz) as can be seen in Fig. 4: N1b at Cz was not determined by the time course of the
S. Bender et al. / NeuroImage 33 (2006) 759–773
Table 3 Significant frontal current sinks during the second frontal N1 peak (Bonferroni correction p < 0.00078) Electrode
Amplitude ± standard deviation; t (p)
Healthy subjects (6–18 years, N = 80) Fz − 81.6 ± 96.0 μV/m2; t = 7.6 F1′ − 26.3 ± 51.6 μV/m2; t = 4.6 F2′ − 24.5 ± 54.1 μV/m2; t = 4.1 AFz′ − 32.2 ± 64.7 μV/m2; t = 4.4
(p < 0.00001) (p = 0.00002) (p = 0.0001) (p = 0.00003)
Headache patients (6–18 years, N = 121) Fz − 92.6 ± 77.2 μV/m2; t = 13.2 (p < 0.00001) F1′ − 28.4 ± 57.8 μV/m2; t = 5.4 (p < 0.00001) F2′ − 27.7 ± 59.9 μV/m2; t = 5.1 (p < 0.00001) AFz′ − 34.9 ± 52.1 μV/m2; t = 7.4 (p < 0.00001)
second frontal peak at Fz but the other way round, central N1b extended to Fz as reflected by a small plateau during N1b in 12– 18-year olds. Spatio-temporal source analysis (a) Grand averages: in good agreement with previous studies (Picton et al., 1995), sources 1, 2, 3 and 4 were fitted reliably for both age groups into the temporal lobe (see Fig. 5). The additional regional source which was used to screen for unexplained activity, was fitted into the frontal lobe reliably only for 12–18 years; for 6– 11 year olds, no systematic unexplained activity was found during N1b. The frontal regional source did not take up any activity in the transversal axis (right-left) indicating that there was no lateralization in frontal activity in our sample. In the age group 12–18 years in the direction towards the vertex (see Fig. 5) there was a peak in the frontal light green dipole’s #5 dipole moment paralleling the central N1b in its time course. This source corresponds to the midfrontocentral current sink in N1b CSD maps. It was missing in 6–11 year olds, supporting the result of CSD analysis that this frontal component during N1b needs maturation before it becomes apparent. In both age groups the frontal regional source showed a second peak (see Fig. 5, dark green dipole #6) with a different orientation about 50 ms after the N1b peak. Its time course paralleled the CSD amplitude at Fz (cf., Fig. 4). So two frontal N1 sources could be dissociated by spatio-temporal source analysis in their maturational development and time course. The midfrontocentral N1b current sink was not produced by temporal overlap of temporal auditory cortex sources with the later frontal peak, instead, two independent frontal sources were found. The “goodness of fit” (percentage of variance explained) indicates if there is considerable variance left which cannot be explained by the model. Residual variance of the model could be reduced to 4.5% in the interval 108–132 ms post S1 (N1b) in the age group 12–18 years introducing the frontal regional source. When the frontal regional source was converted into a single dipole, residual variance increased only to 5.6%, indicating that only one dipole component of the regional source, the one oriented towards the vertex, contributed substantially to the additional variance explained. When the frontal regional source was entirely turned off, residual variance increased to 17.2%. After refitting the temporal dipoles (1, 2, 3, and 4, see Methods section) on the grand average of 6–11-year-old subjects, which produced only small changes in localization and orientation, in this
age group residual variance was 6.0%. When we converted the frontal regional source to a single dipole, this dipole was not oriented towards the vertex. Residual variance increased only to 6.2%, however, when the frontal source was turned off entirely, residual variance was still only 7.3%. So while the frontal regional source explained about 13% of signal variance in the interval 108– 132 ms for 12–18-year olds, for 6–11-year olds this value was reduced to only 1.3%. (b) Individual averages: in order to test for the statistical significance of the development of frontal contributions to N1b, we applied our dipole models with constant locations and individually adjusted orientations to all single subject averages: linear regression of the dipole moment of the frontal N1b source (light green dipole #5 in Fig. 5) showed an increasing frontal contribution to midcentral N1b (24 ms window around the individual peak at Cz) during childhood and adolescence: The regression equation (± standard errors) was 46.6 ± 24.1 μV– 7.5 ± 2.0 μV/year (t = 3.8; p = 0.0003) for healthy subjects and 6.5 ± 15.9 μV– 3.2 ± 1.3 μV/year (t = 2.4; p = 0.02) for headache patients. The negative slope indicates an increase in the contribution to the surface-negative component. The dipole moment (± standard deviation) was highly significant in 12–18-year-old adolescents (− 62.4 ± 52.8 μV; t = 7.6; p < 0.00001 for healthy subjects and − 43.3 ± 41.4 μV; t = 7.4; p < 0.00001 for headache patients) and significantly larger than in 6–11-year-old subjects (healthy children difference − 44.8 μV; t = 3.3; p = 0.001; headache patients difference − 21.7 μV; t = 2.6; p = 0.01). There were no qualitative changes in the temporal dipoles’ activity (see Fig. 5). Please note the temporal coincidence of the second frontal source’s peak (dipole #6) and N1c (dipoles #3 and 4), while P2 occurred later (see the positive deflection in dipole #5 at about 200 ms). Discussion We analyzed the age-related development of N1 components in healthy subjects and primary headache patients. We hypothesized that late maturation of frontal contributions to N1b could lead to qualitative topographic changes in N1b during in childhood and adolescence, indicating qualitative differences in the information processing of children. Moreover, we examined whether this maturation would be disturbed in migraineurs. Our main result was that central N1b amplitude against average reference showed a highly significant continuous increase during childhood and adolescence, when long intertrial intervals in combination with an attention-demanding (CNV) task were used. More than that, this increase was significantly more prominent at Cz than at C3/C4 and resulted in a qualitative change in N1b topography. Topographical analysis of average referenced data, current source density and spatio-temporal source analysis pointed towards a frontal generator which accounted for the maturational changes of N1b (see below). The amount of N1b variance explained by this frontal generator was small in comparison to the potential generators in the temporal lobes and thus it was difficult to detect. However, the frontal generator which produced negativity over the supplementary motor area could be shown in our study in two large independent samples of healthy subjects and headache patients—a fact that allows drawing reliable conclusions. No significant diagnosis-related differences turned out. Most notably, migraineurs showed no maturation disorder with respect to auditory N1. In subjects suffering from tension type headache
S. Bender et al. / NeuroImage 33 (2006) 759–773
maturation was not that evident (see Figs. 3 and 5), however, elevated central N1b amplitudes in young patients depended mainly on a single subject (see Figs. 3 and 5) which proved to be non-significant. The rather long intertrial intervals we used together with the limited number of stimulations in a trial block (both preventing a strong habituation) and the tone burst intensity applied could explain this lack of diagnosis-related differences: habituation and stimulus–intensity dependence but not AEP amplitude have been described to be affected by migraine pathology (Afra et al., 2000; Schoenen et al., 2003; Wang et al., 1996). Differences in the auditory N1 stimulus–intensity slope might be due to diminished amplitudes in migraineurs when stimulated at low intensities rather than excessive orienting at high intensities (Buodo et al., 2004). Increased late CNV over the vertex in migraine children (Bender et al., 2002) could be accounted for by a summation of widespread cortical negativity specifically related to brainstem activation during sustained directed attention during late CNV (Bahra et al., 2001; Diener and May, 1996; Gerber and Schoenen, 1998). In contrast, frontal contributions to N1b seem to reflect different physiological functions (most likely related to a fast communication between auditory and motor areas, allowing a fast reaction due to an orienting response; Naatanen and Picton, 1987) and do not seem disturbed in their maturation in our sample of migraineurs and other primary headache patients. Our findings thus provide further insights into the nature of what is actually disturbed in the information processing of migraineurs (Ambrosini et al., 2003; Buodo et al., 2004) and which functions seem to be unaffected. Furthermore, we could dissociate a second frontal N1 component (Mulert et al., 2001) with a different time course and topography from N1b which could account for an apparent more “frontal” distribution of N1 in children (Ceponiene et al., 2002) when N1b and this peak overlap. This second frontal N1 component showed a very different maturation trajectory during childhood and adolescence and was already present from 6–8 years on. Central N1b development—insertion of the current results into previous findings The increase in negativity during N1b was most prominent over the midfrontocentral area and in good agreement with previous findings (though some methodological differences have to be kept in mind regarding the stimuli, interstimulus intervals and instructions used): most authors have described that N1b increases during maturation, while temporal N1c decreases (Bruneau et al., 1997; Ceponiene et al., 2002; Gomes et al., 2001; Pang and Taylor, 2000; Ponton et al., 2000, 2002). Some authors have failed to find the N1b increase (Goodin et al., 1978; Tonnquist-Uhlen et al., 2003). A shift from a frontal to a central maximum in the topography of midline N1 during adolescence has been a common finding (Bruneau et al., 1997; Ceponiene et al., 2002; Goodin et al., 1978); younger subjects show a more frontal distribution at longer latencies when individual latencies for each channel were used in the analysis (Ceponiene et al., 2002). Our results demonstrate that this might have been the result of an overlap of N1b and a second frontal component peaking about 40 to 50 ms later. The frontal contribution to N1b could depend upon long intervals between consecutive tones (above 8 s in adults) (Hari et al., 1982), as orienting reactions have been described mainly for rare events. The N1b in our study was large and peaked later than usual for stimuli of a rapid onset with shorter intertrial intervals. Source
analysis (see Fig. 5) showed that the frontal N1b source peaked later than the rise of the tangential temporal N1b dipole moments. Thus the increased frontal contribution to adolescents’ N1b could account for longer N1b latency as well as its large amplitude in our study. That there is no significant age-related change in CSD amplitude at C3/C4 (CSD filters out the effect of volume conduction from the temporal lobe) has been shown by Gomes et al. (2001). That N1b amplitude at Cz increases more than the N1b amplitude at C3 or C4 is in good agreement with the findings of Ponton et al. using a monaural stimulation (Ponton et al., 2000, 2002). While specific maturation of a radial N1 component in children has been suggested by comparisons of electrical potentials and magnetic fields (Takeshita et al., 2002), others argue that the apparent movement of the N1 maximum towards more medial locations was due to the decreasing impact of the overlapping lateral N1c component (Gomes et al., 2001). Our data contribute to resolve the controversy about changes in temporal lobe activity or additional frontal sources. The latencies we have found are similar to the latencies in several studies (Ceponiene et al., 2002; Pang and Taylor, 2000), though longer than reported by Gomes et al. (2001) and Tonnquist-Uhlen et al. (2003) most likely reflecting differences in the stimuli used. Short tone bursts have been shown to produce well analyzable N1 complexes (Rosburg et al., 2002; Sams et al., 1985; Vidal et al., 2005). So our data of a specific N1b increase over the vertex during childhood/adolescence and our interpretations of an additional radial frontal source accounting for these maturational differences insert well into previous results. Moreover, our paper is the first detailed description and source analysis of the contributions of different frontal sources to N1b topography changes throughout maturation that has been published so far. Despite the fact that most developmental trajectories might be most adequately modeled using age− 1 (1/age) to account for rapid changes at early ages which afterwards slow down (Bender et al., 2005) (see also Fig. 3), linear regression seemed a sufficient approximation for most parameters except the second frontal N1 peak. Bilateral temporal activity during N1 The projection of tangentially oriented activity in supratemporal plane on central areas (see Fig. 5, red and blue dipoles #1 and #2; Picton et al., 1995; Scherg and Von Cramon, 1985; Vaughan et al., 1980) could be detected reliably already in children at the age of 6– 8 years in our attention-demanding task with long interstimulus intervals. We replicated that N1c activity (most likely arising from radial sources in the temporal cortex (Giard et al., 1994; Picton et al., 1995; Scherg and Von Cramon, 1985), see also Fig. 5, dipoles #3 and #4) decreases with increasing age. There was a certain temporal overlap between N1c and P2. However, P2 peaked later, there were little age effects on P2 at Cz (see Fig. 1) and, though difficult to disentangle, the inversion of P2 over lower leads (see Fig. 1, e.g., leads TP9′, TP10′ at about 200-ms latency) was dissociable from N1c. Anyway, the overlap might have influenced the location of the temporal dipole sources. Frontal N1 component, about 40 to 50 ms after N1b A significant current sink at Fz and surrounding leads (Alcaini et al., 1994) could be shown for all age groups. In contrast to frontal contributions to N1b (possible late maturation of fast
S. Bender et al. / NeuroImage 33 (2006) 759–773
sensory-motor connections/sensory fine-tuning by frontal areas), this second component could be dissociated in its maturational course and is already well developed during childhood. It has been suggested to reflect early response selection processes in the anterior cingulate (ACC) (Mulert et al., 2001). Like early CNV (Bender et al., 2004a,b) or the negative complex (Nc) (Courchesne, 1978; Shibasaki and Miyazaki, 1992) it is prominent already in children from 6 to 8 years on and dissociates from the later development of evoked negativity during motor preparation (Bender et al., 2005; Chiarenza et al., 1983). Dissociation of the maturational trajectories of two frontal components (spatio-temporal dipole source analysis and current source density mapping) A significant independent (clearly separated) current sink at Cz would exclude a vertical projection of temporal activity, because widespread activity from distant sources is filtered out by CSD calculation: a current sink means that the respective area is more negative than its surrounding and thus reflects local activity. It has been shown that no spurious midline current sink is produced by bilateral activity in another context, the Bereitschaftspotential (Knosche et al., 1996). However, spatial resolution was limited even with our high-density sensor array and despite a stronger increase of N1b amplitude at Cz than at C3/C4, a clear separation of the midfrontocentral current sink from the bilateral current sinks around C3 and C4 was not possible because it depended on the interpolation parameters used. Thus we employed dipole source analysis in order to resolve this issue. Spatio-temporal source analysis basically showed that the dipole model of Scherg and Cramon (Picton et al., 1999; Scherg and Von Cramon, 1985; Vaughan et al., 1980) including radial and tangential temporal auditory sources is valid not only for adult subjects but also during development (Ponton et al., 2002). Two bilateral dipole pairs could be fitted temporally for both children (6–11 years) and adolescents (12–18 years) and accounted for most of N1 variance. Sources 1 and 2 reflect tangential activity in the supratemporal plane (primary auditory cortex) during N1b, sources 3 and 4 reflect radial temporal activity of secondary auditory cortex (Giard et al., 1994; Picton et al., 1995). No relevant large-scale differences in the orientation or localization of these dipoles were found during development in 6–11-year-old children. N1b and N1a/N1c showed little temporal overlap but were found to be quite separated in their time course (Giard et al., 1994; Picton et al., 1995), as indicated also by differences in their latencies. Additional frontal sources were necessary to explain the maturational changes in N1b topography: the additional frontal regional source distinguished between two separate frontal activities which differed in time course and maturation: while the frontal source oriented towards the vertex (dipole #5 in Fig. 5) was nearly absent in 6–11year-old children, it was clearly present in 12–18-year-old children. The second frontal source which peaked about 40–50 ms later and was oriented towards frontal areas (dipole #6 in Fig. 5; Alcaini et al., 1994; Mulert et al., 2001) was present in all age groups. Dipoles 1 and 2 were not sufficient to explain adolescent N1b topography, there was no simple addition of negativity towards the midline. The additional explained variance of the frontal source during N1b was about 10–15%. Neglecting it can lead to serious localization errors. Residual variances showed reasonable values. Because grand averages, on which the source models were fitted, comprised subjects from 6–11 and 12–18 years respectively and
maturation was continuous as shown by linear regression analysis, some noise had to be expected despite an overall good signal-tonoise ratio. Most important, the dipole moment of dipole #5 showed a significant maturation in linear regression analysis, reached highly significant values for 12–18-year-old adolescents and indicated that the selective midfrontocentral N1b increase was better explained by additional frontal than by increased bilateral temporal tangential activity. Though the localization of the frontal regional source does not give information about the localization of the frontal generators but has to be viewed as an equivalent dipole, it shows that frontal contributions are necessary to explain N1 (residual variances decreased considerably when the frontal source was introduced) and that dipole source analysis can be used to show the time course of different frontal N1 components. Equivalent dipole orientation (pointing towards the vertex) and corresponding CSD maps (current sink around Cz, FCz) during N1b might point towards activity in the supplementary motor area (SMA) or deeper cingulate motor area (CMA) (Babiloni et al., 1999; Bender et al., 2005; Cui et al., 1999; Naatanen and Picton, 1987). SMA and CMA have been shown to play an important role in early movement preparation (Deecke, 1996). The parallel maturation of the current sinks at Cz during N1b and during movement-related potentials, such as late CNV (Bender et al., 2005), could also indicate related cortical generators. In contrast, equivalent dipole orientation (pointing towards Fz) and CSD maps (current sink around Fz) of the second frontal N1 peak might point towards activation of the anterior cingulate (ACC) (Alcaini et al., 1994; Mulert et al., 2001) which is supposed to be implicated in early response selection and response monitoring (Kiefer et al., 1998; Liddle et al., 2001). What functional developments does the qualitative change in N1 reflect? Influences of frontal cortex on sensory functions have been described (Amassian et al., 1998; Foxe and Simpson, 2002). Transcranial magnetic stimulation of frontal areas influenced visual perception. Thus an increased frontal N1b contribution could reflect maturation of sensory fine-tuning by frontal areas in adolescents. Moreover, our results show, that while some frontal functions which are important for learning processes (response monitoring, most likely linked to ACC; Liddle et al., 2001) might be present already in children (second frontal N1 component), the automatic activation of secondary motor areas (SMA) for fast stereotyped responses, which might facilitate fast reaction times to auditory stimuli in adults, might not be developed yet (midfrontocentral current sink of N1b). Auditory reaction times continue to decrease with increasing age until adolescence, much stronger than central conduction time (Fietzek et al., 2000). Most variability of reaction time is ascribed to the efficiency of the coupling between sensory perception and motor response execution (Menon et al., 1998). Children might react in a qualitatively different way to behaviorrelevant auditory stimuli due to a absent effective short-cut between sensory and secondary motor areas (due to frontal cortex immaturity or missing myelination; Steen et al., 1997). Significant movement-related evoked negativity has been shown in our sample of children over secondary and primary motor areas (Bender et al., 2004a,b, 2005). Thus we think that a different pattern of cortical activation during N1b is a more plausible
S. Bender et al. / NeuroImage 33 (2006) 759–773
explanation for our findings than a delayed micro-structural maturation of (secondary) motor cortex leading to the absence of scalp far field potentials due to less organized dendritic structures (Bender et al., 2005). A complete sensory-motor behavioral cycle from auditory perception to response execution might thus comprise early brainstem responses, first cortical potentials (primary auditory cortex) at about 30–60 ms (Pa, Nb, P50) (Liégeois-Chauvel et al., 1994; Weisser et al., 2001), activation of secondary auditory association areas (N1a) about 75 ms (Giard et al., 1994; Naatanen and Pictn, 1987; Picton et al., 1995), N1b as a correlate of auditory attention (N1b, about 100–120 ms) (Naatanen and Pictn, 1987) including also a coupling from sensory areas and secondary motor areas (SMA) and then •
either fast response execution when stimuli do not have to be classified (simple reaction time paradigm) or early response selection, monitoring/inhibition (second frontal N1 component, ACC, and P2) (Crowley and Colrain, 2004; Mulert et al., 2001) and further stimulus classification (N1c, N2, mismatch negativity) (Gomot et al., 2000; Naatanen, 2003) before a response is executed (Go/Nogo, choice reaction paradigm).
The timing of the possible SMA contribution to N1b (about 120 ms) and the start of lateralized activity (lateralized readiness potential) over premotor and primary motor areas about 100 ms before the button press (Donchin and Coles, 1991; RodriguezFornells et al., 2002) add well to the total simple reaction time of about 220–230 ms in our adolescent subjects (Bender et al., 2005). Trials with erroneous motor responses to the warning stimulus were excluded from the current analysis, thus SMA activation during N1b cannot indicate simple response-related activity in our data. Conclusion In sum, during maturation there was a qualitative change in N1b topography which rather resulted from additional frontal activity overlying the SMA in adolescents than complex changes in temporal auditory cortex activation. In children, such frontal N1b activity was clearly absent in both of our large samples. Thus our data provide first hints, that children might lack an efficient coupling for fast automatic responses without stimulus classification between auditory sensory and secondary motor areas because of late frontal lobe maturation during development. Children might react in a different, less automatic way than adults do. In contrast, early response selection/inhibition in the anterior cingulate might follow a different maturational trajectory and be already present in 6-year-old children. Primary headache patients including migraineurs with and without aura did not show an altered maturation with respect to these parameters. Acknowledgments This work was supported by the Pain Research Programme of the Medical Faculty, University of Heidelberg (F207040, E1). We would like to thank André Rupp (University of Heidelberg) for his remarks on the manuscript.
References Afra, J., Proietti Cecchini, A., Sandor, P.S., Schoenen, J., 2000. Comparison of visual and auditory evoked cortical potentials in migraine patients between attacks. Clin. Neurophysiol. 111, 1124–1129. Alcaini, M., Giard, M.H., Thevenet, M., Pernier, J., 1994. Two separate frontal components in the N1 wave of the human auditory evoked response. Psychophysiology 31, 611–615. Amassian, V., Cracco, R., Maccabee, P., Cracco, J., Rudell, A., Eberle, L., 1998. Transcranial magnetic stimulation in study of the visual pathway. J. Clin. Neurophysiol. 14, 288–304. Ambrosini, A., Rossi, P., De Pasqua, V., Pierelli, F., Schoenen, J., 2003. Lack of habituation causes high intensity dependence of auditory evoked cortical potentials in migraine. Brain 126, 2009–2015. Andersen, V., 2001. Assessing executive functions in children: biological, psychological, and developmental considerations. Pediatr. Rehabil. 4, 119–136. Babiloni, C., Carducci, F., Cincotti, F., Rossini, P.M., Neuper, C., Pfurtscheller, G., Babiloni, F., 1999. Human movement-related potentials vs desynchronization of EEG alpha rhythm: a high-resolution EEG study. NeuroImage 10, 658–665. Bahra, A., Matharu, M.S., Buchel, C., Frackowiak, R.S., Goadsby, P.J., 2001. Brainstem activation specific to migraine headache. Lancet 357, 1016–1017. Bender, S., Weisbrod, M., Just, U., Pfuller, U., Parzer, P., Resch, F., Oelkers-Ax, R., 2002. Lack of age-dependent development of the contingent negative variation (CNV) in migraine children? Cephalalgia 22, 132–136. Bender, S., Oelkers-Ax, R., Resch, F., Weisbrod, M., 2004a. Motor processing after movement execution as revealed by evoked and induced activity. Brain Res. Cogn. Brain Res. 21, 49–58. Bender, S., Weisbrod, M., Resch, F., Oelkers-Ax, R., 2004b. Orienting reaction versus specific response anticipation during early contingent negative variation. Clin. Neurophysiol. 115, 1836–1845. Bender, S., Weisbrod, M., Bornfleth, H., Resch, F., Oelkers-Ax, R., 2005. How do children prepare to react? Imaging maturation of motor preparation and stimulus anticipation by late contingent negative variation. NeuroImage 27, 737–752. Bruneau, N., Roux, S., Guerin, P., Barthelemy, C., Lelord, G., 1997. Temporal prominence of auditory evoked potentials (N1 wave) in 4–8year-old children. Psychophysiology 34, 32–38. Buodo, G., Palomba, D., Sarlo, M., Naccarella, C., Battistella, P.A., 2004. Auditory event-related potentials and reaction times in migraine children. Cephalalgia 24, 554–563. Ceponiene, R., Rinne, T., Naatanen, R., 2002. Maturation of cortical sound processing as indexed by event-related potentials. Clin. Neurophysiol. 113, 870–882. Chiarenza, G.A., Papakostopoulos, D., Giordana, F., Guareschi-Cazzullo, A., 1983. Movement-related brain macropotentials during skilled performances. A developmental study. Electroencephalogr. Clin. Neurophysiol. 56, 373–383. Courchesne, E., 1978. Neurophysiological correlates of cognitive development: changes in long-latency event-related potentials from childhood to adulthood. Electroencephalogr. Clin. Neurophysiol. 45, 468–482. Crowley, K.E., Colrain, I.M., 2004. A review of the evidence for P2 being an independent component process: age, sleep and modality. Clin. Neurophysiol. 115, 732–744. Cui, R.Q., Huter, D., Lang, W., Deecke, L., 1999. NeuroImage of voluntary movement: topography of the Bereitschaftspotential, a 64-channel DC current source density study. NeuroImage 9, 124–134. Deecke, L., 1996. Planning, preparation, execution, and imagery of volitional action. Brain Res. Cogn. Brain Res. 3, 59–64. Diener, H.C., May, A., 1996. New aspects of migraine pathophysiology: lessons learned from positron emission tomography. Curr. Opin. Neurol. 9, 199–201. Donchin, E., Coles, G.H., 1991. While an undergraduate waits. Neuropsychologia 29, 557–569.
S. Bender et al. / NeuroImage 33 (2006) 759–773
Fietzek, U.M., Heinen, F., Berweck, S., Maute, S., Hufschmidt, A., SchulteMonting, J., Lucking, C.H., Korinthenberg, R., 2000. Development of the corticospinal system and hand motor function: central conduction times and motor performance tests. Dev. Med. Child Neurol. 42, 220–227. Fischer, B., Biscaldi, M., Gezeck, S., 1997. On the development of voluntary and reflexive components in human saccade generation. Brain Res. 754, 285–297. Foxe, J.J., Simpson, G.V., 2002. Flow of activation from V1 to frontal cortex in humans. Exp. Brain Res. 142, 139–150. Gerber, W.D., Schoenen, J., 1998. Biobehavioral correlates in migraine: the role of hypersensitivity and information-processing dysfunction. Cephalalgia 18 (Suppl. 21), 5–11. Giard, M.H., Perrin, F., Echallier, J.F., Thevenet, M., Froment, J.C., Pernier, J., 1994. Dissociation of temporal and frontal components in the human auditory N1 wave: a scalp current density and dipole model analysis. Electroencephalogr. Clin. Neurophysiol. 92, 238–252. Gomes, H., Molholm, S., Christodoulou, C., Ritter, W., Cowan, N., 2000. The development of auditory attention in children. Front. Biosci. 5, 108–120. Gomes, H., Dunn, M., Ritter, W., Kurtzberg, D., Brattson, A., Kreuzer, J.A., Vaughan Jr., H.G., 2001. Spatiotemporal maturation of the central and lateral N1 components to tones. Brain Res. Dev. Brain Res. 129, 147–155. Gomot, M., Giard, M.H., Roux, S., Barthelemy, C., Bruneau, N., 2000. Maturation of frontal and temporal components of mismatch negativity (MMN) in children. NeuroReport 11, 3109–3112. Goodin, D., Squires, K., Henderson, B., Starr, A., 1978. Age-related variations in evoked potentials to auditory stimuli in normal human subjects. Electroencephalogr. Clin. Neurophysiol. 44, 447–458. Gratton, G., Coles, M.G., Donchin, E., 1983. A new method for off-line removal of ocular artifact. Electroencephalogr. Clin. Neurophysiol. 55, 468–484. Hari, R., Kaila, K., Katila, T., Tuomisto, T., Varpula, T., 1982. Interstimulus interval dependence of the auditory vertex response and its magnetic counterpart: implications for their neural generation. Electroencephalogr. Clin. Neurophysiol. 54, 561–569. Hegerl, U., Gallinat, J., Mrowinski, D., 1994. Intensity dependence of auditory evoked dipole source activity. Int. J. Psychophysiol. 17, 1–13. Huttenlocher, P.R., 1990. Morphometric study of human cerebral cortex development. Neuropsychologia 28, 517–527. IHS, 1988. Classification and diagnostic criteria for headache disorders, cranial neuralgias and facial pain. Headache Classification Committee of the International Headache Society. Cephalalgia 8, 1–96 (Suppl). Kiefer, M., Marzinzik, F., Weisbrod, M., Scherg, M., Spitzer, M., 1998. The time course of brain activations during response inhibition: evidence from event-related potentials in a go/no go task. NeuroReport 9, 765–770. Knosche, T., Praamstra, P., Stegeman, D., Peters, M., 1996. Linear estimation discriminates midline sources and a motor cortex contribution to the readiness potential. Electroencephalogr. Clin. Neurophysiol. 99, 183–190. Kropp, P., Siniatchkin, M., Stephani, U., Gerber, W.D., 1999. Migraineevidence for a disturbance of cerebral maturation in man? Neurosci. Lett. 276, 181–184. Liddle, P.F., Kiehl, K.A., Smith, A.M., 2001. Event-related fMRI study of response inhibition. Hum. Brain Mapp. 12, 100–109. Liégeois-Chauvel, C., Musolino, A., Badier, J.M., Marquis, P., Chauvel, P., 1994. Evoked potentials recorded from the auditory cortex in man: evaluation and topography of the middle latency components. Electroencephalogr. Clin. Neurophysiol. 92, 204–214. Luciana, M., Nelson, C., 1998. The functional emergence of prefrontallyguided working memory systems in four- to eight-year-old children. Neuropsychologia 36, 273–293. Menon, R., Luknowsky, D., Gati, J., 1998. Mental chronometry using latency-resolved functional MRI. Proc. Natl. Acad. Sci. U.S.A. 95, 10902–10907. Mulert, C., Gallinat, J., Pascual-Marqui, R., Dorn, H., Frick, K.,
Schlattmann, P., Mientus, S., Herrmann, W.M., Winterer, G., 2001. Reduced event-related current density in the anterior cingulate cortex in schizophrenia. NeuroImage 13, 589–600. Muller, B.W., Sartory, G., Tackenberg, A., 2002. The movement-related potential in children with migraine and tension-type headache. Cephalalgia 22, 125–131. Naatanen, R., 2003. Mismatch negativity: clinical research and possible applications. Int. J. Psychophysiol. 48, 179–188. Naatanen, R., Picton, T., 1987. The N1 wave of the human electric and magnetic response to sound: a review and an analysis of the component structure. Psychophysiology 24, 375–425. Nunez, P.L., Pilgreen, K.L., Westdorp, A.F., Law, S.K., Nelson, A.V., 1991. A visual study of surface potentials and Laplacians due to distributed neocortical sources: computer simulations and evoked potentials. Brain Topogr. 4, 151–168. Oelkers-Ax, R., Bender, S., Just, U., Pfuller, U., Parzer, P., Resch, F., Weisbrod, M., 2004. Pattern-reversal visual-evoked potentials in children with migraine and other primary headache: evidence for maturation disorder? Pain 108, 267–275. Oldfield, R.C., 1971. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia 9, 97–113. Pang, E.W., Taylor, M.J., 2000. Tracking the development of the N1 from age 3 to adulthood: an examination of speech and non-speech stimuli. Clin. Neurophysiol. 111, 388–397. Perrin, F., Pernier, J., Bertrand, O., Echallier, J., 1989. Spherical splines for scalp potential and current density mapping. Electroencephalogr. Clin. Neurophysiol. 72, 184–187. Picton, T., Lins, O., Scherg, M., 1995. The recording and analysis of eventrelated potentials. In: Boller, F., Grafman, J. (Eds.), Handbook of Neurophysiology. Elsevier Science. Picton, T.W., Alain, C., Woods, D.L., John, M.S., Scherg, M., Valdes-Sosa, P., Bosch-Bayard, J., Trujillo, N.J., 1999. Intracerebral sources of human auditory-evoked potentials. Audiol. Neuro-Otol. 4, 64–79. Picton, T.W., Bentin, S., Berg, P., Donchin, E., Hillyard, S.A., Johnson Jr., R., Miller, G.A., Ritter, W., Ruchkin, D.S., Rugg, M.D., Taylor, M.J., 2000. Guidelines for using human event-related potentials to study cognition: recording standards and publication criteria. Psychophysiology 37, 127–152. Ponton, C.W., Eggermont, J.J., Kwong, B., Don, M., 2000. Maturation of human central auditory system activity: evidence from multi-channel evoked potentials. Clin. Neurophysiol. 111, 220–236. Ponton, C., Eggermont, J.J., Khosla, D., Kwong, B., Don, M., 2002. Maturation of human central auditory system activity: separating auditory evoked potentials by dipole source modeling. Clin. Neurophysiol. 113, 407–420. Rodriguez-Fornells, A., Kurzbuch, A.R., Munte, T.F., 2002. Time course of error detection and correction in humans: neurophysiological evidence. J. Neurosci. 22, 9990–9996. Rosburg, T., Haueisen, J., Sauer, H., 2002. Stimulus duration influences the dipole location shift within the auditory evoked field component N100m. Brain Topogr. 15, 37–41. Sams, M., Hamalainen, M., Antervo, A., Kaukoranta, E., Reinikainen, K., Hari, R., 1985. Cerebral neuromagnetic responses evoked by short auditory stimuli. Electroencephalogr. Clin. Neurophysiol. 61, 254–266. Scherg, M., Von Cramon, D., 1985. Two bilateral sources of the late AEP as identified by a spatio-temporal dipole model. Electroencephalogr. Clin. Neurophysiol. 62, 32–44. Schoenen, J., Ambrosini, A., Sandor, P.S., Maertens de Noordhout, A., 2003. Evoked potentials and transcranial magnetic stimulation in migraine: published data and viewpoint on their pathophysiologic significance. Clin. Neurophysiol. 114, 955–972. Sheehan, K.A., McArthur, G.M., Bishop, D.V., 2005. Is discrimination training necessary to cause changes in the P2 auditory event-related brain potential to speech sounds? Cogn. Brain Res. 25, 547–553. Shibasaki, H., Miyazaki, M., 1992. Event-related potential studies in infants and children. J. Clin. Neurophysiol. 9, 408–418.
S. Bender et al. / NeuroImage 33 (2006) 759–773 Siniatchkin, M., Kropp, P., Gerber, W.D., Stephani, U., 2000. Migraine in childhood—Are periodically occurring migraine attacks related to dynamic changes of cortical information processing? Neurosci. Lett. 279, 1–4. Steen, R.G., Ogg, R.J., Reddick, W.E., Kingsley, P.B., 1997. Age-related changes in the pediatric brain: quantitative MR evidence of maturational changes during adolescence. Am. J. Neuroradiol. 18, 819–828. Takeshita, K., Nagamine, T., Thuy, D., Satow, T., Matsuhashi, M., Yamamoto, J., Takayama, M., Fujiwara, N., Shibasaki, H., 2002. Maturational change of parallel auditory processing in school aged children revealed by simultaneous recording of magnetic and electric cortical responses. Clin. Neurophysiol. 113, 1470–1484. Tonnquist-Uhlen, I., Ponton, C., Eggermont, J., Kwong, B., Don, M., 2003.
Maturation of human central auditory system activity: the T-complex. Clin. Neurophysiol. 114, 685–701. Vaughan Jr., H.G., Ritter, W., Simson, R., 1980. Topographic analysis of auditory event-related potentials. Prog. Brain Res. 54, 279–285. Vidal, J., Bonnet-Brilhault, F., Roux, S., Bruneau, N., 2005. Auditory evoked potentials to tones and syllables in adults: evidence of specific influence on N250 wave. Neurosci. Lett. 378, 145–149. Wang, W., Timsit-Berthier, M., Schoenen, J., 1996. Intensity dependence of auditory evoked potentials in migraine: an indication of cortical potentiation and low serotonergic neurotransmission? Neurology 46, 1404–1409. Weisser, R., Weisbrod, M., Roehrig, M., Rupp, A., Schroeder, J., Scherg, M., 2001. Is frontal lobe involved in the generation of auditory evoked P50? NeuroReport 12, 3303–3307.