Differential recruitment of high frequency wavelets (600 Hz) and primary cortical response (N20) in human median nerve somatosensory evoked potentials

Neuroscience Letters 256 (1998) 101–104

Differential recruitment of high frequency wavelets (600 Hz) and primary cortical response (N20) in human median nerve somatosensory evoked potentials Fabian Klostermann*, Guido Nolte, Florian Losch, Gabriel Curio Neurophysics Group, Department of Neurology, Klinikum Benjamin Franklin, Freie Universita¨t, Hindenburgdamm 30, 12200 Berlin, Germany Received 5 August 1998; received in revised form 18 September 1998; accepted 19 September 1998

Abstract Human median nerve somatosensory evoked potentials contain a burst of high-frequency (600 Hz) wavelets superimposed on the primary cortical response (N20). These presumably reflect highly-synchronized repetitive thalamic and/or intracortical population spike bursts and are diminished in non-REM sleep with N20 persisting. Here the burst/N20 relation in awake subjects was examined by using eight different intensities of electric median nerve stimuli. In all subjects the amplitude recruitment of both N20 and burst could be modeled adequately as a sigmoidal function of stimulus intensity. While 8/10 subjects showed a parallel recruitment, 2/10 subjects required significantly higher stimulation intensities for burst than for N20 recruitment. This dampened burst recruitment possibly reflects slight vigilance fluctuations in open-eyed awake subjects; a further increase of burst thresholds could explain the burst attenuation when entering shallow sleep.  1998 Elsevier Science Ireland Ltd. All rights reserved

Keywords: Somatosensory system; Median nerve; Primary cortical response; High frequency burst; Recruitment

Human somatosensory evoked potentials (SEP) elicited by median nerve stimulation show a parietal negativity peaking at 20 ms (N20), commonly interpreted as the primary cortical response which is generated by excitatory postsynaptic potentials (EPSP) in apical dendrites of pyramidal cells in Brodmann area 3b [1]. In addition, high-frequency wavelets have been described as superimposed on the ascending limb of N20 [3–10,13,14,20]. Since the main energy of the wavelet burst is at 600 Hz [6] it can be isolated from the underlying low-frequency N20 using digital highpass filtering [8]. Thalamic and/or cortical generators have been discussed to contribute to these wavelets and to reflect mainly the timing of highly synchronized repetitive population spikes, e.g. in thalamocortical afferences or in postsynaptic intracortical cell populations [5]. EEG [9,20] as well as MEG data [14] showed a distinct attenuation of these high frequency wavelets in non-REM sleep states while N20 remained unaffected or even in* Corresponding author. Tel.: +49 30 84452002; fax: +49 30 84454264; e-mail: [email protected]

creased [14]. Here, a possible dissociability of primary cortical response and burst activity during wakefulness was investigated by analyzing the recruitment of N20 and 600 Hz activity as function of stimulus intensity in awake subjects. Median nerve SEP were recorded in 10 healthy subjects lying in a supine position (7 males, 3 females; 24–39 years). They were instructed to stay awake with their eyes open in a brightly lit room and to pay attention to the stimulus; to stabilize wakefulness at a level as high as possible subjects were interviewed, e.g. about sleep tendencies, every 2 min after each run. Transcutaneous electrostimulation was applied to the median nerve at the right wrist (0.1 ms constant-current squarewave pulses; 18.3 Hz; 2000 stimuli/run). SEP were recorded from C3′ to F3 (frontal ground electrode; impedances below 5 kQ; 5–1500 Hz hardware filtering). Off-line digital highpass filtering (corner frequency: 428 Hz) was performed for isolation of high frequency potentials. In each subject eight runs were recorded: stimulation intensities were varied from subthreshold values for elicit-

0304-3940/98/$19.00  1998 Elsevier Science Ireland Ltd. All rights reserved PII S0304- 3940(98) 00773- 3

102

F. Klostermann et al. / Neuroscience Letters 256 (1998) 101–104

ing either N20 or burst response up to fully recruited responses in a randomized order to avoid habituation effects. Intensity steps of 6.25% or 12.5% of motor threshold (MT) were used (maximal range across all subjects: 37.5% to 150% of MT). The accuracy of stimulus intensities was within 0.1 mA. For burst analysis two response intervals were chosen: the first one (width about 8 ms) was centered around the maximum burst amplitude between 16 to 24 ms and contained burst and noise; the second interval from 30 to 45 ms (width 15 ms) did not contain any burst signal and was used to provide a valid estimate of the noise level. The precise width of the first interval was defined individually at maximal burst recruitment; since burst peak latencies were found similar at the different stimulation intensities this interval included the whole burst in all eight runs. An estimate of the noise power (derived from the second, later interval) was subtracted from the power in the burst interval. The square root of this difference is proportional to the burst amplitude. N20 peak amplitude (in a bandpass 5–428 Hz) was measured from a sloping baseline which was linearly extrapolated from the underlying slow late SEP component in this steady state stimulation paradigm. To quantify the signal amplitude as a function of stimulus intensity a sigmoidal model was employed. In this model each neuron contributing to the burst or N20 has a threshold stimulus intensity (I) above which it is recruited to fire; the neuronal thresholds are Gaussian distributed as function of I around a mean threshold intensity I0. The amplitude of the evoked response, which to first approximation is proportional to the number of firing neurons, can be found by integrating the Gaussian function. The resulting error-function was fitted to burst amplitudes and, respectively, N20 amplitudes as function of stimulus intensity in each subject.

A dissociation in recruitment would be indicated by different intensities (I0) required for crossing the 50%-amplitude level. In all ten subjects both N20 and burst could be identified. The rms-value for the fully recruited burst was 0.06 ± 0.04 mV (mean across subjects ± SD). On average, bursts began to rise above noise at 16.72 ± 0.67 ms, and fell below at 22.28 ± 1.34 ms. The average duration of the burst was 5.56 ± 1.8 ms. The average peak latency for a fully recruited N20 was 19.97 ± 0.68 ms; the average burst/N20 amplitude ratio was 1:18. The goodness of fit (explained variance) for the N20 error function was 97.1 ± 1.9%, and 95 ± 5.5% for the burst. In eight subjects the recruitment of N20 and burst response was not significantly different (minimal P . 0.15; Fig. 1A and 2A). In contrast, in two subjects a highly significant (Student’s t-test, two-tailed; P , 0.0001 and P , 0.0004) dissociated recruitment was found (Fig. 1B): their burst response required higher stimulus intensities than the N20 response for an equal percentage of full amplitude to be recruited (Fig. 2B); both subjects showed regular sigmoidal recruitment patterns and saturation amplitudes for both N20 and burst responses close to the group average. Accordingly, a parallel recruitment of N20 and burst appears as a regular physiological behaviour in awake subjects; however, the highly significant dissociation of response recruitments in a minor group of subjects must be integrated when discussing possible burst generator mechanisms. These widely divergent response patterns observed in this unselected group of subjects document the possibility of uncoupling burst and N20 generation also in wakefulness and allow for two alternative (state vs. trait) interpretations. (a) The recruitment patterns may reflect intraindividually

Fig. 1. Stack plot of SEP recordings (10–40 ms after median nerve stimulus) rearranged in order of increasing intensity (1–8) for two characteristic subjects. In subject (A) N20 and 600 Hz activity were recruited in parallel, in subject (B) the recruitment of 600 Hz activity started at a higher intensity level than N20 recruitment (bandpass for N20: 5–428 Hz; burst: 428–1500 Hz).

F. Klostermann et al. / Neuroscience Letters 256 (1998) 101–104

fixed but interindividually variable ‘trait’ properties, i.e. the two different recruitment patterns of burst and N20 in the awake state are ideosyncratic, stable functions of stimulus intensity and might indicate independent threshold processes. This model, however, would not encompass the burst/N20 dissociation in sleep found regularly in all subjects [9,14,20]. (b) While the N20 recruitment appears to be relatively stable, the burst recruitment may be variable even between psychophysiologically just slightly different awake ‘states’: the burst recruitment might be dampened (i.e. it would require higher stimulus intensities) due to a subject’s less efficient continuous stabilization of high-level vigilance; this could have occurred despite the open-eye instruction and the interviews repeated at two minute intervals. Such a variable burst recruitment explains a possible response dissociation in awake state as well as the absence of the burst in non-REM sleep; the latter could then be interpreted as a pronounced shift of burst recruitment towards higher stimulation intensities which could not be realized without waking up the subject. Concerning loci of possible burst generators both MEG and EEG studies have accumulated evidence for minimally two (one cortical, one subcortical) contributions: cortically, a close colocalization with the N20 source at area 3b was found [6,14]; while these superficial burst generators were shown to follow the somatotopy in S–I [7], it was not yet possible to discriminate possible contributions from repetitive discharges carried in thalamocortical axon terminals and/or postsynaptic intracortically generated spiking. Subcortical burst sources were described in deep brain recordings from the thalamocortical radiation in awake human patients [15] and have been found to contribute also to non-invasive EEG/MEG recordings [4,10]. A burst recruitment at stimulus intensities higher than required for the primary cortical postsynaptic response (N20) could indi-

103

cate a higher threshold in thalamocortical projection cells for a sharply synchronized burst output than for asynchronous single spike output. A subset of cells in ventrobasal thalamus can discharge with intraburst frequencies of about 600 Hz [16]. Mostly, thalamic neurons burst on the basis of low-threshold (LTS) calcium spikes leading to a burst of fast high-frequency sodium spikes riding on the LTS crown (for review see Ref. [17]). Notably, however, these bursts occur mainly in sleep and disappear during arousal in contrast to the SEP 600 Hz bursts which were found maximal during arousal and attenuated during sleep. Furthermore, a subset of thalamocortical projection cells in the visual thalamus of awake cats showed high-frequency bursts which were interpreted to act as a modality-specific alerting mechanism [12]. Hence, it appears that the invasively [15] and non-invasively [4,10] detected subcortical SEP bursts are not related to thalamic LTS spikes. At present, however, alternatives cannot be excluded, e.g. that the synchronization between bursting thalamic LTS cells is less precisely stimulus-locked in sleep (possibly due to the strong rhythmic background activities) than in wakefulness [19] so that they just would not show up after the massive averaging which is required for obtaining the scalp SEP response. Pyramidal cells in primary cortices and fast-spiking inhibitory GABA-ergic interneurons can discharge in bursts up to 800 Hz [11,18], and could add to the burst generation [5,6,13,14] provided they would discharge highly phase locked to the stimulus. Also intracortical sequential synaptic activity, as described e.g. for the inter-layer (IV–II) synaptic impulse propagation in visual cortex [2], might contribute to the burst. A burst recruitment at higher stimulus intensities than required for the primary postsynaptic cortical response (N20) could be easily reconciled with all these secondary (intracortically generated) responses. In conclusion, high-frequency SEP wavelets allow the non-invasive derivation of a parameter related to the timing

Fig. 2. Different recruitment patterns of N20 and 600 Hz response (same representative subjects as in Fig. 1). (A) The fit of the error function to the data shows a parallel recruitment, while in (B) it is dissociated with a shift of 600 Hz response to higher intensities than required for the recruitment of N20 response. The saturation amplitude of each signal was normalized to 1. The error bars correspond to the noise level computed from the residues of the fit.

104

F. Klostermann et al. / Neuroscience Letters 256 (1998) 101–104

of repetitive population spikes of thalamic and/or cortical origin; in some subjects this high frequency burst can be dissociated from the cortical primary postsynaptic N20 response by means of variations in stimulus intensity in the awake state. Further analyses of functional aspects of burst responses may allow the integration of cellular firing properties and macroscopic network states like ‘vigilance’ or ‘attention’.

[10]

[11]

[12] [1] Allison, T., McCarthy, G., Wood, C.C. and Jones, S.J., Potentials evoked in human and monkey cerebral cortex by stimulation of the median nerve: a review of scalp and intracranial recordings, Brain, 114 (1991) 2465–2503. [2] Bode-Greuel, K.M., Singer, W. and Aldenhoff, J.B., A current source density analysis of field potentials evoked in slices of visual cortex, Exp. Brain Res., 69 (1987) 213–219. [3] Curio, G., High frequency (600 Hz) bursts of spike-like activities generated in the human cerebral somatosensory system, Electroenceph. clin. Neurophysiol., Suppl. (1998) in press. [4] Curio, G., Losch, F. and Nolte, G., Magnetoencephalographic detection of deep thalamocortical activity evoked by median nerve stimulation, Electroenceph. clin. Neurophysiol., 106 (1998) 40P. [5] Curio, G., Mackert, B.-M., Abraham-Fuchs, K. and Ha¨rer, W., High-frequency activity (600 Hz) evoked in the human primary somatosensory cortex: a survey of electric and magnetic recordings. In C. Pantev, Th. Elbert and B. Lu¨tkenho¨hner (Eds.), Oscillatory Event-Related Brain Dynamics: NATO ASI Series A: Life Sciences, Vol. 271, Plenum Press, New York, 1994, pp. 205–218. [6] Curio, G., Mackert, B.-M., Burghoff, M., Koetitz, R., AbrahamFuchs, K. and Ha¨rer, W., Localization of evoked neuromagnetic 600 Hz activity in the cerebral somatosensory system, Electroenceph. clin. Neurophysiol., 91 (1994) 483–487. [7] Curio, G., Mackert, B.-M., Burghoff, M., Neumann, J., Nolte, G., Scherg, M. and Marx, P., Somatotopic source arrangement of 600 Hz oscillatory magnetic fields at the human primary somatosensory hand cortex, Neurosci. Lett., 234 (1997) 131–134. [8] Eisen, A., Roberts, K., Low, M., Hoirch, M. and Lawrence, P., Questions regarding the sequential neural generator theory of the somatosensory evoked potential raised by digital filtering, Electroenceph. clin. Neurophysiol., 59 (1984) 388–395. [9] Emerson, R., Sgro, J.A., Pedley, T.A. and Hauser, W.A., Statedependent changes in the N20 component of the median nerve

[13]

[14]

[15]

[16]

[17] [18]

[19]

[20]

somatosensory evoked potential, Neurology, 38 (1988) 64– 68. Gobbele´, R., Buchner, H. and Curio, G., High-frequency (600 Hz) SEP activities originating in the subcortical and cortical human somatosensory system, Electroenceph. clin. Neurophysiol., 108 (1998) 182–189. Gray, C.M. and McCormick, D.A., Chattering cells: superficial pyramidal neurons contributing to the generation of synchronous oscillations in the visual cortex, Science, 274 (1996) 109–113. Guido, W. and Weyand, T., Burst responses in thalamic relay cells of the awake behaving cat, J. Neurophysiol., 74 (1995) 1782–1786. Hashimoto, I., Physiological mechanisms and functional significance of high-frequency oscillations from the human somatosensory cortex. In I. Hashimoto and R. Kakigi (Eds.), Recent Advances in Human Neurophysiology. International Congress Series, Excerpta Medica, Elsevier, Amsterdam, 1998, in press. Hashimoto, I., Mashiko, T. and Imada, T., Somatic evoked highfrequency magnetic oscillations reflect activity of inhibitory interneurons in the somatosensory cortex, Electroenceph. clin. Neurophysiol., 100 (1996) 189–203. Katayama, Y. and Tsubokawa, T., Somatosensory evoked potentials from the thalamic sensory relay nucleus (VPL) in humans: correlations with short latency somatosensory evoked potentials at the scalp, Electroenceph. clin. Neurophysiol., 68 (1987) 187–201. Rasmusson, D.D., Changes in the response properties of neurons in the ventroposterior lateral thalamic nucleus of the raccoon after peripheral deafferentation, J. Neurophysiol., 75 (1996) 2441–2450. Sherman, S.M. and Guillery, R.W., Functional organization of thalamocortical relays, J. Neurophysiol., 76 (1996) 1367–1395. Swadlow, H.A., Beeloozerova, I.N. and Sirota, M.G., Sharp local synchrony among putative feed-forward inhibitory interneurons of rabbit somatosensory cortex, J. Neurophysiol., 79 (1998) 567–582. Timofeev, I., Contreras, D. and Steriade, M., Synaptic responsiveness of cortical and thalamic neurones during various phases of slow sleep oscillation in cat, J. Physiol., 494 (1996) 265–278. Yamada, T., Kameyami, S., Fuchigami, Y., Nakazumi, Y., Dickins, Q.S. and Kimura, J., Changes of short latency somatosensory potentials in sleep, Electroenceph. clin. Neurophysiol., 70 (1988) 126–136.