Cohen Veterans Bioscience (CVB), a non-profit research biotech advancing brain health solutions, today announces findings from a study which generates new evidence in support of a critical brain imaging biomarker, that may help guide people who suffer from post-traumatic stress disorder (PTSD) or major depressive disorder (MDD) towards the most effective treatment.
The study, entitled "Identification of psychiatric disorder subtypes from functional connectivity patterns in resting-state electroencephalography (EEG)" was led by Stanford University professor of psychiatry and behavioral sciences, Amit Etkin, MD, PhD, and received major funding from CVB.
In the findings, reported in the journal https://www.nature.com/articles/s41551-020-00614-8, the authors used advanced machine learning techniques on data from high density resting-state EEG signals to identify robust and distinct functional connectivity patterns in brain circuits enabling subtyping of patients, independent of clinical diagnosis.
Currently, there are no tools to predict treatment outcomes for either PTSD or MDD patients. Although psychotherapy is considered the most efficacious approach for the PTSD population generally, many individual patients fail to respond to this treatment. Similarly, antidepressant medications improve outcomes only modestly over placebo in MDD patients. The failure of these patient groups to respond to treatment may be due to the fact that the clinical diagnosis of these neuropsychiatric disorders involves biologically heterogeneous populations who respond differently to treatment.
Dr. Etkin is taking these finding into development for clinical use as founder of the startup company Alto Neurosciences.
There is a great need in psychiatry for objective tests that can inform diagnostic development and clinical treatment decisions for heterogeneous conditions such as PTSD and MDD. Our findings are exciting because they reflect progress towards identifying evidence-based biomarkers, and they also demonstrate the value of machine learning techniques for advancing a personalized approach to treatment – which are part of a tipping point in the field."
Amit Etkin, MD, PhD, Founder of Alto Neurosciences
In the study, the patient subtypes found in PTSD and MDD did not differ in terms of clinical symptom severity prior to treatment, but had different responses to treatments. These subtypes were identified based on functional connectivity patterns, or neural signatures, found through electroencephalography (EEG), a low-cost method for quantifying brain function that can be done in the context of clinical care. Interestingly, researchers identified similarly strong subtype-related connectivity differences in PTSD as well as in MDD patients. One of the subtypes responded poorly to either psychotherapy or antidepressant medications. However, both subtypes responded similarly to noninvasive transcranial magnetic brain stimulation (TMS) treatment. This suggests that, for one subtype, psychotherapy or medication treatment is best, while for the other subtype they may be better served by advancing faster to TMS treatment – all based on a "biological brain diagnosis" with EEG that is independent of traditional clinical criteria for PTSD or MDD.
These discoveries have significant implications as they help stratify individuals independent of clinical diagnosis based on what may represent a new transdiagnostic biomarker. This will enable discovery of a new generation of precision therapeutic discoveries and targeted treatments."
Dr. Andreas Jeromin, Chief Scientific Officer, CVB
In this study, researchers applied machine learning to interpret signals from EEG," adds Lee Lancashire, PhD, Chief Information Officer at CVB. "This technique is interesting in that it jointly identifies the most important features in the data while clustering the participants into similar groups in a data-driven and unbiased way, not incorporating any prior knowledge about the clinical presentation of PTSD or MDD. The fact that EEG-based subpopulations identified in PTSD also replicated in MDD is a step forward in understanding these heterogeneous disorders, and further highlights the power of machine learning."
Zhang, Y., et al. (2020) Identification of psychiatric disorder subtypes from functional connectivity patterns in resting-state electroencephalography. Nature Biomedical Engineering. doi.org/10.1038/s41551-020-00614-8.