Psychiatric disorders, like most other nervous system disorders, lack biomarkers in clinical use. Instead, diagnosis of psychiatric disorders rests on patients’ reports of their symptoms, signs from their mental status examination, and clinician observations of their behavior. To make a diagnosis, mental health professionals group those clues into distinct diagnostic categories listed in one of two classification systems, DSM-IV and ICD-10 (American Psychiatric Association, 2000; World Health Organization, 2007).The categories listed there are based on expert consensus that draws from both scientific evidence and clinical experience. The diagnostic categories are largely descriptive in orientation, with DSM actively professed to be “neutral with respect to theories of etiology” (American Psychiatric Association, 2000). But should the diagnostic categories of psychiatric disorder drive the search for biomarkers? Are there complementary alternatives to using standard diagnostic classifications? Those were the provocative questions raised by Dr. Steven Hyman, provost of Harvard University.

Growing evidence suggests that biomarker research might best be served by focusing elsewhere. Hyman proposed that biomarker research should focus less on current categories of disorder and more on underlying clinical states for which some knowledge of pathophysiology or neurocircuitry is available. Clinical states of this kind often transcend the boundaries of a single category of disorder. For example, the cognitive impairment observed in schizophrenia (including impairment of working memory) is associated with thinning of prefrontal cortex observed by structural MRI (Hyman, 2007b). Although it is responsible for substantial disability, it is not part of the DSM-IV criteria, which date to earlier understandings of schizophrenia as primarily reflecting psychotic symptoms such as hallucinations and delusions. A focus on biomarkers to follow working memory deficits involving prefrontal cortical circuits would seem more likely to succeed than searching for a biomarker of DSM-IV schizophrenia, which is a heterogeneous syndrome defined only by symptoms and course.

To understand the change in emphasis, Hyman first traced the intellectual and historical underpinnings of the current diagnostic classification systems and then argued that excessive reliance on current, consensus diagnostic categories—especially for the purpose of biomarker research—may lead researchers down blind alleys.