Discovering informative biomarkers in psychiatry.

Arguably, there is not a more important theme in psychiatry today than biomarkers. We don’t have many; some would say, none. The panel of authors of Abi-Dargham et al's paper1 is eminent and broad, and they have been capable to summarize our current needs and advise on developing and using biomarkers in our field. The authors note a gap in the biomarker validation process, i.e., in demonstrating that each individual biomarker is valid, reliable and useful. They are certainly correct in this respect. For diseases of the cardiovascular and metabolic systems, we have lots of biomarkers. Some of them were identified before they became useful, so that, once ready for validation, the data were already collected to analyze. In psychiatry, we still have a good deal of data collection to do. An exceptional model for advancing biomarker discovery is represented by the Framingham Heart Study, a project which was motivated by President Roosevelt's untimely death from cardiovascular disease in 1945 and has been ongoing since 19482. We have untold thousands of discoveries, pieces of data and follow-on innovations in cardiology all deriving in part from this rich, longitudinal, trans-generational, deep phenotyping, cohort study. It has been both the scientific organization and the structure of the effective collaborations which has enabled this project to transform knowledge in cardiology. Also, the idea of the learning health care system3 was introduced to help clinicians conceptualize how to use the health care system to not only deliver care, but to associate clinical facts and biomarkers to outcomes in order to advance knowledge in the field. These are two models useful in anyone's compendium on biomarkers. But, it is the complexity of unexplained neural biology that makes the work harder and slower in psychiatry. Within this framework, there are certain principles describing biomarker development within our field that seem secure, several of which are articulated in Abi-Dargham et al's paper. In addition, in psychiatry, the field needs to discover and use biomarkers of brain function, not only of psychiatric diagnoses. It needs to be said that we do not even know if the field has its diagnoses correct, with skepticism in this area fueled by polygenicity and pleiotropy across many disease traits4. Psychiatric diagnoses are not biologically founded, and can lead to imprecise conclusions. The field most critically needs brain biomarkers associated with characteristics of disease constructs: their risk factors, mechanisms, treatment, course and outcomes. The idea of a true cohort study is to use the variability in large subject samples to define true patterns of the expression of a biomarker within a brain disorder, rather than cripple the clinical outcomes with biased samples. We may not be accustomed to the necessary large size of such studies2, and they may be costly. Costly, however, in the short run, with payoff in their application. It is important to notice and avoid uncritical research on biomarkers which result, not from the illness, but from the chronic effects of drugs used to treat the illness. Biomarkers around dopamine functions in psychosis and noradrenergic functions in depression fall into this category, unless the trial design allows treatment vs. disease to be distinguished. It is true that sometimes the strongest hypotheses develop from therapeutic categories, e.g. research on dopamine in schizophrenia, but their validation must involve the use of strategies such as recruitment of drug-free subjects, animal studies of chronic drug effects and/or testing the biomarker on and off from treatment drugs. In addition, the idea of applying a biomarker battery, i.e., collecting multiple diverse measures from a target population and clustering them within a single population, is frequently productive. Often, it is too difficult to anticipate or hypothesize which biomarker may really be useful for what, without using a battery of them, as the first step. Application of a full biomarker battery in a particular disorder is informative with respect to what biomarkers cluster with each other and which cluster with a target outcome, using replication to build confidence in an interpretation. Indeed, in the Bipolar and Schizophrenia Network for Intermediate Phenotypes (BSNIP) study5, it was the entire biomarker battery which went into defining the final characteristics of the experimental biomarker-defined entities, which were called “psychosis biotypes”. These biotype constructs have provided research with an alternative to phenomenologically defined entities, as a stage in developing final disease categories. Moreover, it is the full biomarker battery which can be applied to distinguishing and understanding defined features of the illness, such as negative symptoms6. BSNIP researchers have developed several individual studies, now ongoing, to test the clinical applicability of the above biotype constructs. One such study tests the hypothesis that biotype 1, with its low intrinsic EEG activity, is a biomarker which indicates responsiveness to clozapine; specifically, we test the hypothesis that increasing intrinsic EEG activity with clozapine in biotype 1 will correlate with symptomatological improvement, using the attractor network model6. A second study, designed to predict treatment response in early psychosis, hypothesizes that the biotypes will define good (biotype 3), moderate (biotype 2) or poor (biotype 1) response to standard coordinated specialty care (CSC)6. In each of these examples, a double-blind trial of the biomarker observation (now ongoing) is necessary, and its application can only be supported if this is done with rigorous design. There is no doubt that considerable hard work will have to go into the study of biomarkers in psychiatry before we are able to bring them to a clinically useful place. Yet, the validation of biomarkers, as reviewed in Abi-Dargham et al's paper, can be so decisive for the future of our field that these studies need to be conducted. Costs have to be born. Yes, wisely; but urgently.