man during waking states, but can be applied to sleep as well. For the hippocampal theta recordings (Fig. 2), power spectra have been used to perform statistical comparisons on extracted features under different behavioral conditions; dominant frequency associated with resting amounts values of 4 Hz which shift to about 5.5 Hz during walking. Rat EEG-recordings to relate beta-increases to midazolam plasma levels mentioned above are another good application of power spectral analysis. As pointed out, pharmacologist derive drug potency from dose (► ED50) or concentra-tion-EEG effect relationship; for animals treated with ► flumazenil, a blocker of the benzodiazepine receptors, higher ► midazolam concentrations are necessary to induce the same EEG changes, which fits with a competitive agonist-antagonist interaction model. ► Benzodiazepines are widely studied because of known properties as anxio-lytic, anticonvulsant, muscle relaxant, and sedative-hypnotic effects, but other mechanisms and therapeutic application can benefit from EEG. Straightforward examples of applied neuropharmacology in animal EEG are seen, for example, in natural or experimentally induced epileptiform, activity of great interest in the understanding and development of new ► anticonvulsants or novel compounds for the induction of ► general anesthesia. The translatability to humans is straightforward and is illustrated in the next section.
Shortly after computerized EEG analysis became available, the idea of using EEG as a diagnostic tool in patients suffering from neurological/psychiatric disorders began to be explored. EEG recording is harmless, easy to implement, and of low cost. Computational methods are available, but need group clusterization, hence are not very useful for individual subjects. EEG is commonly used in the diagnosis of epilepsy, sleep disorders, and as a research tool in a number of CNS-disorders. Cognitive decline is associated with EEG changes: a hallmark in Alzheimer's patients is the increase in slow waves and the decrease in alpha activity. Basic knowledge on the progression rate in EEG deteriorations needs follow-up tests over weeks or months. This can be assessed on different occasions in eyes-closed recordings which last for only a few minutes. An example of the magnitude (approaching EEG-amplitude, see Kiebel et al. 2005) as a function of frequency or spectrogram is shown in Fig. 4 for a healthy subject; the recording lasted 3 min.
Finally, the technique is widely applied for on-line monitoring of anesthesia. An illustration of benzodiazepine sedation can be found in Fig. 5a. It is important to stress the similarity in time course of both plasma levels
0 10 20 30
0 10 20 30
Electroencephalography. Fig. 4. Example of signal representation in the frequency domain as a spectrogram. The shaded areas give demarcations between classically defined bands; eyes-closed resting EEG can easily be repeated, has good test-retest reliability (not shown) and is recommended in drug studies. y-axis: square-root of power per spectral value yield a "magnitude," the center of gravity (=median frequency on x-axis) is normally around 9-10 Hz. Inset: Bargraph with labels of extracted features according to the names of the frequency bands. Drugs can almost independently modulate these, and this can even be different depending on the topographic location.
(plateau around 30 min) and the drop in EEG overall frequency (following a transient elevation in line with increased fast frequencies). Compartmental modelisation of PK-PD relationships fully validates the causality between input (drug i.v.) and output seen on the EEG. Sub-anesthetic doses of midazolam for the volunteer of Fig. 5b, for which EEG modifications are less pronounced than in the first case, rely on the same mechanistic principle and yield a good EEG pharmacodynamic response timing in relation to infusion onset. As for beta-EEG in animals shown previously, proper consensus on model characteristics is provided, e.g., ► triazolam (Greenblatt et al. 1994) for high frequency content in humans as is the case for midazolam (Fiset et al. 1995) but on the content of the whole spectrum.
Signals as shown in left panel from Fig. 3a and "raw" spectrogram of Fig. 4 can be used to evidence a range of modulations/perturbations. Thus, the level of vigilance is reflected by the spectral composition of the brain signals.
Electroencephalography. Fig. 5. (a) Pharmacokinetics for a single injection of the "midazolam-like" drug Ro 48-6791 at controlled infusion rate (3 ng/mL/min) in induction of anesthesia and conscious sedation. Curve peaks around 30 min when subject stopped responding. (c) changes in median frequency (qEEG); note the dramatic drop to 1-2 Hz center of gravity which remains stable around 30 min (adapted from Ihmsen H, Albrecht S, Hering W, Schüttler J, Schwilden H (2004) Modeling acute tolerance to the EEG effect of two benzodiazepines. Br J Clin Pharmacol 57(2):153-161). (b) Pharmacokinetic profile in a separate (own) experiment for a single i.v. injection for 15 min of a nonanaesthetic dose of midazolam (5 mg, 1/15th of dose for full sedation); note the rapid rise and fall in mother compound, and delay in (alpha-hydroxy) metabolite concentrations (concentration difference in the order of about one log-unit). (d) off-line analysis of the same parameter as in (c) but the lowering is less than one third of predrug values.
Was this article helpful?