October 23, 2013. The release of the latest version of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) earlier this year kicked off a round of public debate about how psychiatric disorders should be classified, though the discussion in the clinical and research communities has been going on for several years (see SRF related news story; SRF Live Discussion). The World Congress of Psychiatric Genetics kicked off on October 18, 2013, with a morning plenary session titled “Defining Mental Illness Through Genetics,” wherein panelists from a broad range of backgrounds discussed the role of genetics in this classification, or nosology, of mental illnesses.
Moderator Steven Hyman from the Broad Institute just across the Charles River in Cambridge, Massachusetts, set the stage by outlining the challenges faced by the psychiatric genetics field. Hyman emphasized that the DSM and its global counterpart, the International Classification of Diseases (ICD), are not psychiatric “bibles”—as they are so often pegged in the media—but rather medical classifications. As cognitive schema imposed on data in order to make them useful, classifications are frequently complicated and messy, he said. So it’s not necessarily surprising that the current categorical system doesn’t map well onto psychiatric genetics (see SRF related news story).
In a preview of the talks to come, Hyman noted that “disorders that turn out to be heterogeneous and polygenic [like mental illnesses] are often better mapped either as spectra or as quantitative dimensions.” Given that so many genes contribute to psychiatric illness, he doubted that genetics would ever be used on its own to diagnose, but suggested that it could be used to identify useful biomarkers.
Historical and epidemiological perspectives
The first panelist was Ken Kendler of Virginia Commonwealth University in Richmond, who provided a historical framework for how genetics has influenced the classification of psychiatric disorders. Kendler, a self-proclaimed “DSM warrior,” described six phases in the relationship between nosology and genetics, beginning with early clinicians who took family history into account when assigning diagnoses. Subsequent "traditional" genetic epidemiological methods of family and twin studies and model fitting have given way—in the molecular genetic era—to candidate gene studies, genomewide association studies, and, most recently, polygenic analyses, he said. The phases have yielded three major levels of nosologically relevant genetic data: familial aggregation/heritability, single variants and aggregates of variants, and polygene scores.
Kendler emphasized that genetics has played an important and increasing role in psychiatric nosology, saying, “I don’t think this shows any signs of changing.” A critical question moving forward, he suggested, is for nosologists to determine what they want from genetics, adding that it depends on the model of psychiatric illness used. For syndromal diagnoses like DSM, nosologists generally want aggregate data, he said, so polygene scores are more useful than single variants. However, the move “toward a more etiologically based diagnosis shifts the picture,” said Kendler, “and the kind of information we get from the single variants could potentially be one of a variety of inferences” that help to clarify the disease mechanisms.
Myrna Weissman of New York’s Columbia University introduced herself as the token epidemiologist of the group. She called for the inclusion of more somatic diseases into the phenotypes of psychiatric illnesses and the consideration of pleiotropy, in which a single gene affects multiple phenotypic traits. The action of genes extends below the neck to affect the rest of the body, she noted, citing the genetic association of interstitial cystitis—a chronic inflammation of the bladder—with panic disorder as an example.
“Epidemiology is an observational science,” said Weissman, which makes it hard to address underlying disease mechanisms. She went on to describe several ways in which epidemiology can inform disease pathophysiology. Existing large, representative sample sizes can be used to generate novel phenotypes. An example of this approach is a recent study in the famous Dunedin cohort (see SRF related news story) that proposes a general “p factor” that marks risk for psychopathology (Caspi et al., 2013, in press). Longitudinal, developmental designs can help to determine the earliest phenotypes. In addition, family studies as well as those conducted on high-risk subjects can identify endophenotypes that are independent of disease expression, said Weissman. Finally, the effect of environmental exposures can be examined through birth cohort studies. “However, for any of these observations to be successful,” she cautioned, “they need to be linked to clinical and basic science approaches.”
Echoing Weissman, Jan Buitelaar from Radboud University Nijmegen in the Netherlands called for a reconceptualization of mental illnesses as diseases of both the brain and body. Another issue, he said, is that the psychiatric genetics field has been focused too much on identifying main gene effects. While important, these effects are sparse and relatively small. A better approach, he said, is to identify more complicated gene interactions, both with other genes and with the environment.
Buitelaar also praised the efforts to date. “So far, genetics has brought huge conceptual innovations to psychiatry,” he said, pointing to the current notion of schizophrenia as a synaptic disease. He also distinguished between two different meanings of the word “definition” in psychiatry. One meaning is concerned with describing the essence and underlying mechanisms of an illness, while the other pertains to delineating and describing the boundaries between disorders. “I think genetics is a very powerful entry into identifying these disease mechanisms at the molecular/cellular level and the neural systems level,” he said. On the other hand, he predicted that genetics would not be useful in further delineating current nosological categories (the second definition).
Buitelaar suggested that uncovering disease mechanisms will require investigations of functionality, rather than a sole focus on structure variants. In line with this, he also emphasized the importance of integrating genetics with other systems such as immunology and energy metabolism, and reiterated the need to study gene-by-environment interactions.
In contrast to Buitelaar, Michael Owen of Cardiff University in the United Kingdom focused on a slightly different use of the word “definition”: how clinicians could use genetics to diagnose mental illnesses. He first turned to the finding that genes operate across diagnostic boundaries. While Owen had expected genetic overlap between schizophrenia and bipolar disorder, he was surprised that it extended into childhood psychiatric disorders such as intellectual disability. Like earlier speakers, Owen also highlighted the pleiotropy of psychiatric illnesses.
Continuing a theme of the session, Owen commented, “I think it’s unlikely that genetic effects are going to map onto clinically useful phenotypes, though this is still a matter for empirical inquiry.” He suggested that genetic variation may be more closely related to illness course, outcome, and response to treatment—areas that need further research.
In that case, genetics may play a role, but as “part of a multilevel psychiatric diagnosis” alongside other modalities of information such as imaging, where it could be used to target treatment and possibly to indicate prognosis.
A major goal of future genetics will be to try to identify both pathogenic and protective mechanisms, an important source of clues to new interventions, he said. Only by understanding the biology will we be able to solve the problems of nosology, he added. “But clearly there’s a long way to go.”
Classifications of the future
Owen acknowledged that the DSM is the best available approach, but added, “where it’s really toxic is if we let it drive our research.” On that note, the session wrapped up with a highly anticipated presentation from Bruce Cuthbert of the National Institute of Mental Health.
“We have all been serving as a prefiguration for the final panelist,” said Hyman as Cuthbert approached the microphone. Cuthbert detailed his institute’s recent undertaking to re-conceptualize mental illness manifestations for research—an effort termed the Research Domain Criteria (RDoC) project. The earlier presentations in the session focused on the need to use genetics, neuroscience, and behavioral science to inform etiology—strategies that will move toward a “true precision approach to diagnosis and treatment,” he said. However, a major roadblock of this personalized medicine is that the current standards and, indeed, “the whole machinery of our system,” are based on DSM diagnoses.
The goal of RDoC is to move beyond the DSM to classify psychiatric illnesses based on “dimensions of observable behavior … and neurobiological measures that implement these kinds of behaviors,” explained Cuthbert. The new framework is based on five domains of functioning, each consisting of component constructs that can be measured at various levels of analysis: genes, molecules, cells, circuits, physiology, behaviors, and self-reports. But the current matrix is not the final version, said Cuthbert. It’s really just a beginning framework. “We have to go back to ground zero and figure out a new way to start, and this looks like a reasonable starting place,” he added. New constructs will likely be added in the future as research progresses.
Cuthbert also emphasized that, beyond a conceptual framework, RDoC is also a research grant funding mechanism. Although the project was introduced slowly, NIMH in the future will be increasingly funding RDoC-based studies over ones that use traditional DSM diagnoses. He concluded by outlining the implications of RDoC for genetic studies. There will be a move toward studying the genetics of complex traits that cut across diagnostic boundaries, he said, an approach that will require new sampling methods and research designs.—Allison A. Curley.