9 January 2006. The venerable electroencephalograph (EEG) has undergone some refinement since the German scientist Hans Berger first recorded human brain electrical activity from his young son's brain in the 1920s. Arrays of electrodes are now monitored, and the resultant data managed by powerful computer software, but the basic procedure hasn't changed much over the years: A small soft electrode is attached to the head with a sticky, highly conductive substance and the activity of countless neurons is generated as wave forms, whether scratched out on paper by a pen or rendered in bright colors on a monitor. When subjects are presented with sensory stimuli, the EEG shows event-related potentials (ERPs)—changes in the wave forms that are not always attributable to any particular neurons or structures in the brain, but which are nonetheless valuable for their consistency and, it is argued, ability to point out underlying abnormalities.
There is consistent evidence that certain EEG and other electrophysiological measures of brain activity are abnormal in schizophrenia patients, as well as in some of their first-degree relatives. Combining some of these measurements may enhance their use as endophenotypes for the purposes of genetic studies, argues a paper published online last December 17 in Biological Psychiatry. Greg Price and colleagues at the University of Western Australia in Perth and the University of Newcastle, Callaghan, Australia, report that combining components of EEG waveforms such as mismatch negativity, P50, P300, along with antisaccade responses, creates an electrophysiological endophenotype with greater value for research than any of these features alone.
Various research groups have found alterations in ERP components in people with schizophrenia. With the further finding that first-degree relatives of people with schizophrenia also show abnormal ERP components, their value as intermediate or "endophenotypes" has been enhanced. Irving Gottesman, at the University of Minnesota, Minneapolis, and colleagues have argued that endophenotypes may be "closer" to any underlying genetic abnormalities than the diagnostic phenotypes of the disorder such as psychosis or flattened affect (Gottesman and Gould, 2003). (For a great deal more reading on this topic, see Gottesman's preliminary text and commentary by other researchers for SRF's discussion of endophenotypes in psychosis research.)
Following the suggestion of William Iocono of the University of Minnesota (Iocono, 1998), Price and colleagues decided to examine whether they could get a greater correlation of schizophrenia-related ERP abnormalities with the disorder by combining four tests:
1. The mismatch negativity (MMN) response, which is a component of the ERP generated about 200 milliseconds after the brain detects a sound deviating from a familiar test pattern (an "oddball" paradigm). This is reduced in amplitude in people with schizophrenia and first-degree relatives.
2. The P50 (for an ERP waveform seen 50 microseconds post-stimulus), which is proposed to be a measure of auditory gating in the brain, a reflection of the fact that the brain pays less attention to the second of two identical sound stimuli (see SRF related news story). Normally, the P50 to the second sound is much lower than that to the first, but in people with schizophrenia and their first-degree relatives, the response is barely diminished.
3. The P300 ERP component, which is an increase seen about 300 milliseconds after a deviant tone that the subject has been asked to listen for in an oddball paradigm. Reduced amplitude of the increase has been reported in people with schizophrenia and first-degree relatives.
4. Antisaccades, which are eye movements away from a visual stimulus that suddenly appears in the visual field. Overriding the natural impulse to look at the new stimulus is more difficult for people with schizophrenia, as well as their close relatives.
Price and colleagues replicated the findings for these individual endophenotypes in their study subjects with schizophrenia (n = 60) versus controls (n = 44), and for MMN and antisaccades among family members of the probands (n = 53). However, the authors report that they were surprised to find that, with the exception of the antisaccade measures, there was little correlation among the different endophenotypes (similar to findings in schizophrenia by Javitt et al., 1995 and Louchart de la Chappelle et al., 2005, but in contrast to Cadenhead et al., 2002).
A multivariate model of all four endophenotypes, using data from just probands and controls (excluding family members of the probands), was 80 percent accurate in distinguishing between the two groups, a better performance than any of the endophenotypes alone. When the researchers reassigned proband family members into proband and control groups, where "proband" now signifies the presence of the abnormal electrophysiologic responses whether or not schizophrenia is present, they found a greater degree of group separation in the multivariate model. "This multivariate endophenotype might be used to increase the power in genetic linkage and association analyses," the authors conclude.—Hakon Heimer.
Price GW, Michie PT, Johnston J, Innes-Brown H, Kent A, Clissa P, Jablensky AV. A multivariate electrophysiological endophenotype, from a unitary cohort, shows greater research utility than any single feature in the Western Australian Family Study of Schizophrenia. Biol Psychiatry. 2005 Dec 17; [Epub ahead of print] Abstract