Using genetics to improve traditional psychiatric diagnoses

Psychiatry has begun the laborious effort of preparing the DSM-V, the new iteration of its diagnostic manual. In so doing, it once again wrestles with the task set by Carl Linnaeus, to "cleave nature at its joints." However, these "joints," the boundaries between psychiatric disorders, such as that between bipolar disorder and schizophrenia, are far from clear. Prior versions of DSM followed the path outlined by Emil Kraeplin in separating these disorders into distinct categories. Yet, we now know that symptoms of bipolar disorder may be seen in patients with schizophrenia and the reverse is true, as well. Further, our certainty about the boundary of these disorders is undermined by growing evidence that both schizophrenia and bipolar disorder emerge, in part, from the cumulative impact of a large number of risk genes, each of which conveys a relatively small component of the vulnerability to these disorders. And since many versions of these genes appear to contribute vulnerability to both disorders, the study of common gene variations has raised the possibility that there may be diagnostic, prognostic, and therapeutic meaning embedded in the high degree of variability in the clinical presentations of patients with each disorder. In addition, many symptoms of schizophrenia and bipolar disorder are traits that are present in the healthy population but are more exaggerated in patient populations. To borrow from Einstein, who struggled to reconcile the wave and particle features of light, our psychiatric diagnoses behave like waves (i.e., spectra of clinical presentations) and particles (traditional categorical diagnoses). Although new genetic approaches may revise our current thinking, such as studies of microdeletions, microinsertions, and microtranslocations of the genome, the wave/particle approach to psychiatric diagnosis places a premium on understanding the "real" clustering of patients into subtypes as opposed to groups created to correspond to the current DSM-IV.

Latent class analysis is one statistical approach for estimating the clustering of subjects into groups. In their study of 270 Irish families, published in the July 15th issue of Biological Psychiatry, Fanous and colleagues conducted this type of analysis, with subjects clustered into the following groups: bipolar, schizoaffective, mania, schizomania, deficit syndrome, and core schizophrenia. When they divided the affected individuals in the study using this approach, they found four regions of the chromosome that were linked to the risk for these syndromes that were not implicated when subjects were categorized according to DSM-IV diagnoses. Dr. Fanous notes that this finding "suggests that schizophrenia as we currently define it may in fact represent more than one genetic subtype, or disease process." According to John H. Krystal, M.D., Editor of Biological Psychiatry and affiliated with both Yale University School of Medicine and the VA Connecticut Healthcare System: "Their findings advance the hypothesis that the variability in the clinical presentation of patients diagnosed using DSM-IV categories is meaningful, providing information that may be useful as DSM-V is prepared. However, we do not yet know whether the categories generated by this latent class analysis will generalize to other populations." This paper highlights an important aspect of the complexity of establishing valid psychiatric diagnoses using a framework adopted from traditional categorical models.

Source: Elsevier