Biomarkers for clinical use in psychiatry: where are we and will we ever get there?

Almost all aspects of psychiatric practice currently rely on assessing the presence and change in symptoms to diagnose and manage patients. Psychiatric disorders are diagnosed based on clusters of symptoms occurring together for at least a minimum period of time, as defined in the DSM-5 and ICD-11. The efficacy of new treatments for psychiatric disorders, and the approval of new medications by regulatory authorities, rely only on changes in symptom severity based on rating scales. However, most treatments for psychiatric disorders are effective only for about half of patients and, without any predictive tools to guide treatment decisions, the interventions offered to any given patient are typically based on clinician and patient preferences. Given this unsatisfactory state of affairs, it is clear that psychiatry, more than any other specialty in medicine, needs clinically useful predictive biomarkers to advance diagnosis and treatment of patients. So, what are biomarkers and how could they help? The US Food and Drug Administration - National Institutes of Health Biomarker Working Group defines a biomarker as “a characteristic that is objectively measured and evaluated as an indicator of normal biologic processes, pathogenic processes, or responses to an exposure or an intervention”. Based on the clinical applications, biomarkers can be classified into: diagnostic biomarkers which aid in the detection of a disease; susceptibility/risk biomarkers for predicting the risk of development of a disease; predictive biomarkers which predict response or non-response to an intervention; monitoring biomarkers which indicate the change in the status of a disease; prognostic biomarkers which aid in prediction of remission or recurrence; and safety biomarkers which predict the likelihood of an adverse event following an intervention. Biomarkers are used widely to aid in the diagnosis and management of diseases in many medical and surgical specialties. For example, before the discovery of biomarkers, Alzheimer's disease (AD) diagnosis was primarily based on the clinical symptom profile, as the definitive diagnosis required post-mortem brain pathology. The diagnostic process was transformed with the discovery of imaging and cerebrospinal fluid biomarkers which can now be used to confirm the diagnosis of AD in living humans1. Given the urgent and pressing need for biomarkers to transform psychiatric practice, the state of the field review of candidate biomarkers in psychiatry by Abi-Dargham et al2 is most timely. They correctly point out that the “litmus test for biomarkers in psychiatric disorders is their ability to change clinical practice”. While their review identifies some promising biomarker candidates for further testing, sadly none has gone through all the stages of validation required for biomarker development, and few (if any) hold the promise of meaningful sensitivity and specificity for adoption in clinical practice. Thus, their conclusion that “we do not yet have clinically actionable biomarkers in psychiatry” is fully warranted. Needless to say, despite decades of significant investments in biomarker research, the lack of progress in discovering clinically useful biomarkers for psychiatry is disappointing. Abi-Dargham et al2 discuss fundamental barriers to biomarker research in psychiatry, including excessive reliance on case-control study designs, heterogeneity of psychiatric disorders, insufficient knowledge of the brain mechanisms and functioning, and confounding effects of age, sex and medication status. Indeed, study designs comparing patients with DSM diagnoses vs. healthy controls have yet to find precise neurobiological/neurochemical alterations underlying symptom expression of psychiatric disorders, a major impediment for targeted discovery of biomarkers. This is not surprising, given the heterogeneity of many DSM-defined psychiatric disorders, as one would not expect the same underlying biological alterations in diverse subgroups of patients. The difficulties in defining the “appropriate phenotype” for biomarker discovery and validation are further compounded by the poor inter-rater agreement for various DSM diagnoses3 and the presence of comorbidities, medication effects and chronicity amongst other factors. Furthermore, despite rapid advances in imaging to study structure, connectivity, neurochemicals and their receptors, and functioning of brain, methods to explore several processes occurring at cellular and molecular levels in the brain are not yet feasible. While animal models have been developed for many psychiatric disorders, none meets the triad of face validity, construct validity and predictive validity, thus limiting their utility in providing neural insights into these conditions. For these reasons, our ability to gain a full understanding of neurobiological and neurochemical alterations in brains of people with psychiatric disorders remains very limited. Given these challenges, will we ever see biomarkers that are relevant for clinical use in psychiatry? Abi-Dargham et al2 offer some suggestions for advancing biomarker discovery, such as focussing on promising biomarker candidates identified in their review, designing studies with an explicit goal of discovering biomarkers for a particular indication, embracing alternate forms of classification for testing potential biomarkers in subgroups of patients based on neurobiological features, adequately powered epi/genetic studies of mega-samples well characterized in clinical course and treatment response, and a priori stratification approaches to test preventive and therapeutic approaches. These are all useful avenues to pursue for biomarker research. In addition, advances in the use of human-induced pluripotent stem cell (iPSC) technology4, especially the iPSC-based three-dimensional (3D) tissue engineering as an in vitro model for diseases5 and CRISPR-Cas9 gene editing, should be leveraged to interrogate and understand molecular mechanisms underlying psychiatric disorders in order to facilitate biomarker discovery. As well, standard data collection protocols should be developed for deep clinical phenotyping, cognitive assessments, biological sampling, and electrophysiological and imaging procedures, to enable pooling of data from centers around the world. The AD Neuroimaging Initiative (ADNI) is an exemplar of such effort6. ADNI began in 2004 with substantial public-private partnership funding that allowed academic centers internationally to standardize data collection and pool data, which led to discovery of biomarkers for AD. Similar initiatives in psychiatry, such as the Establishing Moderators and Biosignatures of Antidepressant Response for Clinical Care for Depression (EMBARC) project, the Canadian Biomarker Integration Network in Depression (CAN-BIND), the Personalized Prognostic Tools for Early Psychosis Management (PRONIA) Consortium, and the planned longitudinal cohort study by the recently launched BD2 Integrated Network7, are clearly steps in the right direction. Moreover, industry-sponsored phase 2/3 clinical trial programs that ascertain the efficacy of new drugs for psychiatric disorders generate vast amounts of treatment data. These data could be a huge resource for biomarker discovery if the trials implement standardized data collection protocols that include deep clinical phenotyping and biological sampling, and the data are made available for pooling with other networks. Looking to the future, the probability of discovering diagnostic biomarkers that map precisely to specific DSM-5 disorders is very low, given the heterogeneity of the disorders and the symptom overlap among them. However, the emerging evidence reviewed by Abi-Dargham et al and the continuing advances in research methods for biomarker discovery offer a ray of hope that susceptibility markers for disease conversion and predictive biomarkers for treatment response will become a future reality in psychiatry.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.