H.H. Stassen, G. Bomben and D. Hell
Human brain wave patterns (EEG) are complex traits as is illustrated by the fact that the mode of inheritance is not a simple mendelian one for the great majority of parameters used to quantify EEG characteristics. One exception may be the low-voltage EEG that is characterized by the absence of an alpha rhythm in the resting EEG and is hypothesized to be an autosomal-dominant trait [Vogel 1986; Anokhin et al. 1992; Steinlein et al. 1992a, b]. Although univariate EEG-parameter models do not readily yield insights into the mode of inheritance of EEG characteristics, there is presumptive evidence from studies on monozygotic (MZ) and dizygotic (DZ) twins, that the inter-individual variability of the human EEG under "normal" conditions is largely genetically determined [Stassen 2003]. In fact, quantitative analysis of the genetic component of the human EEG on the basis of twins brought up together and reared apart [Stassen et al. 1988a, b] yielded conclusive proof that the variance in individual resting EEG patterns is predominantly determined by heritable factors: we found a heritability coefficient h²=0.825 for adolescents ranging in age from 9 to 12 years (reared together), a heritability coefficient h²=0.761 for adults between 20 and 35 years old (reared together), and a heritability coefficient h²=0.719 for adults aged 21 to 65 years (reared apart). The corresponding quotients rDZ/rMZ of within-pair DZ and within-pair MZ similarities were with 0.499, 0.452 and 0.598 not too far from the theoretical value of 0.5 of the polygenic-additive model.
Our interest in the present study of a family consisting of 8 sisters and 4 brothers has a psychiatric background. We conjecture that investigations into the heredity of brain wave patterns may contribute to a better understanding of the complex familial aggregations of psychic illnesses. As compared to psychic illnesses, brain wave patterns exhibit a similar complexity on the trait level but on the other hand, they are "objectively" assessable, i.e. in a rater-independent way and in a standardized experimental setting. We therefore designed this study of a family with 8 sisters and 4 brothers in order to quantify the intra-familial phenotypic heterogeneity of EEG characteristics and to investigate the question of whether EEG characteristics are quantitative or qualitative traits. To this end, each individual was assessed twice at an interval of 14 days so that the reproducibility of results could be tested and the between-sib differences could be qualified with respect to the stability of the underlying parameters over time. Specifically, we addressed the following questions: (1) To what degree do EEG parameters posses trait-like qualities? (2) Are EEG parameters associated with qualitative or quantitative traits? (3) If quantitative, what is the distribution in the general population: unimodal? normally distributed? (4) Is the between-sib EEG similarity half the within-pair EEG similarity of monozygotic twins?
Our family sample consisted of 12 healthy sibs (8 sisters, 4 brothers) raised in the same environment and ranging in age between 30 and 48 years. The probands were asked to fill out the "Zurich General Health" (ZGF) questionnaire which comprises 80 items and addresses lifetime consumption behavior, somatic disorders, psychosomatic disturbances and psychic impairment. For each of these probands EEGs were recorded twice at 14-day intervals under the condition of quiet wakefulness (eyes closed) using standard 10-20 electroencephalic lead placements. Each recording comprised 5 minutes of resting EEG derived from 6 parallel channels (P3-O1, P4-O2, T3-T5, T4-T6, T5-O1, T6-O2). All signals were digitized online at a sampling rate of 256 Hz and with a resolution of 12 bits. Subsequently, the resulting time series were subdivided into epochs of 20 second length, plotted, and inspected visually for artifacts. Intervals identified within the 20-second epochs as disturbed by artifacts (e.g. by body movements, eye blinks, muscle potentials, loose electrodes, etc.) were marked with an artifact code according to our standards.
Most EEG parameters possess distinct trait-like qualities as indicated by a high specificity (informativeness) and by a high stability over time. Accordingly, these parameters were deemed to represent a sound basis for our investigation into the between-sib EEG similarities. Detailed analysis revealed a considerable between-sib variation. For example, with respect to "absolute alpha-band power" the 12-sib sample was partitioned into 5 subgroups: a subgroup of 3 sibs with parameter values around 180; a subgroup of 5 sibs with parameter values around 280; a subgroup of 2 sibs with parameter values around 380; a subgroup of 1 sib with a parameter value around 480; and a subgroup of 1 sib with a parameter value around 580. In terms of the variation observed in the general population, the variation found in this sib sample lay in the lower part of the distribution of the general population and covered nearly 50% of the total range. And this picture of high within-family variation was essentially the same for all parameters under consideration.
In our multivariate approach to between-sib EEG similarity we used 2 different types of feature vectors: (1) vectors that encompassed parameters taken from the frequency bands 3.75-7.5 Hz, 7.5-15 Hz and 15-30 Hz. These vectors were compared by means of the Euclidean distance, and (2) 256-dimensional spectral patterns that were compared by means of a set-theoretical similarity measure. The analysis yielded a significantly higher between-sib similarity (p<0.0001) in comparison to the between-subject similarity of unrelated individuals. On the other hand, the between-sib similarity was significantly lower (p<0.0001) than the within-subject similarity derived from repeated assessments at 14-day intervals, and also significantly lower (p<0.0001) than the within-pair similarity of monozygotic twins.