Abstract
Analysis of previously published CSF monoamine data has revealed statistical evidence for a biological subtype of depression, characterized by abnormally low CSF-5HIAA and CSF HVA. Using maximum likelihood gaussian mixture analysis we were able to resolve the empirical frequency distributions of both CSF HVA and CSF 5-HIAA into two component normal mixtures. Simultaneous analysis of both CSF 5-HIAA and CSF HVA revealed a two component bivariate normal mixture distribution in which 35% of the depressed patient sample were classified in the low subgroup. No evidence for mixture distributions was found in controls. Analysis of CSF MHPG revealed a single component normal distribution with virtually identical mean and variance in both patients and controls. These results are shown to be virtually identical to parallel analyses conducted on CSF monoamine data collected as part of the NIMH collaborative study on the psychobiology of depression study.