Previous studies have shown that changes in brain functionClinicians long have observed a lag time of several weeks between the initiation of antidepressant treatment and clinical response for many patients (Hyman and Nestler 1996;Katz et al. 1996). Some individuals do have early symptomatic improvement, and this has been reported to predict further improvement over the next several weeks (Nierenberg et al. 1995). Reports have suggested that some physiologic changes are seen shortly after initiation of treatment (Sulser 1989;Beck 1995;Dahmen et al. 1997). No clinically practical physiologic predictor of treatment response has yet been identified with these techniques, however, and the relationship of early physiologic changes to eventual clinical outcome remains incompletely understood.Quantitative electroencephalography (QEEG) has been used as a physiologic measure in efforts to address these questions. Prior work with "pharmaco-EEG" techniques has shown that the administration of antidepressant compounds yields reproducible changes in EEG activity in healthy control subjects within a few hours of dosing (Saletu et al. 1982(Saletu et al. , 1983(Saletu et al. , 1986(Saletu et al. , 1987(Saletu et al. , 1987 Grunberger 1985, 1988;Sannita et al. 1983;Sannita 1990;
Early Prefrontal Changes in Depression 121Itil et al. 1984;Herrmann et al. 1991;Luthringer et al. 1996). The relationship of these immediate EEG changes in control subjects to eventual clinical response in a depressed population is unclear. Other QEEG work with depressed subjects has found that changes from baseline in theta power early in the course of treatment may characterize groups of depressed patients who are more likely to respond to antidepressant treatment (Ulrich et al. 1994). Unfortunately, the overlap in the value of these changes between responder and nonresponder groups precluded the use of this measure in response prediction for individual subjects, and prior research did not indicate how to relate changes in theta power to other measures of regional brain activity (e.g., regional cerebral blood flow or metabolism). We previously have shown that absolute and relative power are complementary measures of brain activity (Leuchter et al. 1993). A relatively new QEEG measure, "cordance," combines information from both absolute and relative power measures (Leuchter et al. 1994a(Leuchter et al. , 1994b. The algorithm yields two indicators: a categorical value ("concordant" or "discordant" state) and a numerical value for each electrode. In an earlier report with the categorical measure , we observed that depressed subjects exhibiting the concordant state prior to treatment had better treatment outcomes when treated with fluoxetine than did subjects with the discordant state. In this report, we use the num*erical values of cordance, because they allow examination of changes in regional brain activity with treatment. In validation against data collected simultaneously with [H 2
15O]-positron emission tomography (PET), cordance values in the the...