Developing a psychiatric numerical taxonomy
Discriminant functions were developed to serve as an objective neurobiological adjunct to psychiatric diagnosis. EEG records were collected from large cohorts of patients in each of the following diagnostic categories: children with ADHD or learning disabilities, adults with Major Affective Disorders, Obsessive Compulsive Disorder, Schizophrenia, chronic substance abuse, elderly patients with subjective as well as objective complaints of cognitive impairment, and mild traumatic head injury. Using QEEG or ERP variables, it proved possible to define stepwise multiple discriminant functions that could replicably separate patients in each of these categories from an age-matched group of normal healthy control subjects.
The fact that discriminant functions using only QEEG variables can accurately separate patients in many DSM diagnostic categories from normals might occasionally serve as a useful adjunct to reinforce diagnoses based upon symptomatology. Diagnosis alone does not necessarily provide a reliable guide to treatment, since individuals who share the same diagnosis may respond very differently to a particular treatment. In order to use QEEG to develop a numerical taxonomy, large groups of patients considered homogeneous with respect to diagnostic classification have been subjected to subtyping using cluster analysis based upon QEEG variables. Different subtypes thus identified have different electrophysiological profiles, implying different pathophysiologies that might display responsiveness to different treatments or display different evolution of illness.
The validity of this approach has been demonstrated by our reported and replicated success in identifying several subtypes among groups of patients diagnosed with Attention Deficit Disorder with and without Hyperactivity, Lupus Erythymatosis, Major Affective Disorder, Drug Dependence and Obsessive-Compulsive Disorder, which displayed different treatment outcomes or progression of illness.
In early 2007 we published a multicenter study showing six QEEG subtypes within a large cohort of patients with a variety of formal diagnoses all, of whom displayed psychotic symptoms. Each subtype contained individuals with a wide variety of DSM IV diagnoses. In ongoing studies of a cohort of patients diagnosed with Autistic Spectrum Disorders and another cohort with Chronic Refractory Pain, we have identified a number of QEEG subtypes. Exploratory studies seeking methods to treat such patients differentially in a more individualized prescriptive manner are in progress.
The Brain Research Laboratories has established and maintained the largest database in the world, with over 15,000 recordings of EEG and Event-Related Potentials from normal subjects and patients with a wide variety of developmental, neurological and psychiatric disorders. In previous work, over 1000 patients in eight different diagnostic categories were subjected to cluster analysis based upon a set of QEEG variables selected by the criterion of their heterogeneity of variance across this cohort. A replicable structure of 12 clusters or subtypes of pathophysiology was identified and successfully replicated. Every cluster contained individuals from several DSM-IV categories and the representatives of each of the 8 categories in the cohort were distributed among at least several of the clusters.
This study demonstrated that a taxonomy based upon neurobiological features can be established. Given the QEEG functional brain imaging and MRI/DTI/MRS technology now available, this offers the possibility of systematically parsing the population of patients with developmental, neurological and psychiatric disorders into a limited number of subtypes with different electrophysiological patterns of pathophysiology, localizing within the brain the Regions of Interest (ROI's) of most salient abnormal function, studying the connectivity among these ROIs using diffusion tensor imaging and investigating the neurochemical and neurotransmitter imbalances implicated in these dysfunctions using magnetic resonance spectroscopy. Such studies, while undoubtedly requiring much time and effort, should enable great strides to be made in more effective treatment of these disorders.