Precision Psychiatry and the Mental Health Treatment Revolution

“On a scale of one to ten, how are you feeling today?”

That question—typically accompanied by a series of “smiley faces” that mutate from sad to happy—and others relating to appetite, sleep, mood, energy, concentration, and more are often the first thing a patient has to address when checking in at an appointment with a psychiatrist. This subjective, semi-quantitative, self-reported measurement is a severe oversimplification of a patient’s mental illness; it is also not the most useful information for a clinician and pales in comparison to the cutting-edge technologies for collecting and analyzing data that are revolutionizing other indications such as cancer.

“Seeing patients in the oncology wards was a profoundly gratifying and rewarding opportunity,” Jacob Donoghue, MD, PhD, reminiscing on his experience upon leaving MIT and returning to Harvard Medical School for the clinical portion of his physician-scientist training, told Inside Precision Medicine. “Machine learning tools with next-generation biotechnologies meant that we were giving truly curative therapies to patients. To be able to radically change the course of kids’ lives, suffering from late-stage leukemia and lymphoma, it was just amazing. Then I got to the neurology and psychiatry rotations, and I felt like I was going back 100 years in time. It didn’t look anything like it looked [in oncology], where we’re giving engineered T cells.”

Donoghue’s observation is not an isolated experience—it is repeated time and time again by clinicians and researchers working in psychiatry. Yet, of all the medical fields, psychiatry has arguably the most to gain from becoming rooted in “precision medicine,” not just because it is behind other areas of discipline but because of the as yet unraveled complexity of the human brain and mind. In the past decade, there has been a push for the development of tools for prognosing, diagnosing, and treating mental illness to usher in a new era of psychiatry: the era of “precision psychiatry.”

Predicting future suicide attempts

The synergy between next-generation sequencing (NGS) and well-characterized biology has been revolutionary in bringing precision medicine to oncology. Many cancer cases come down to a single mutation in a single gene, resulting in unregulated cellular proliferation. Today in oncology, aggregating germline variants using meta-features that capture their genomic, nucleotide, and epigenetic contexts can improve cancer risk prediction.

But that has not been the case in psychiatry, and not because it has not been looked into. Countless studies and resources have gone into finding the genetic causes of psychiatric illnesses. While there has been progress, there has yet to be a situation where the malfunction of a single gene is sufficient to explain a specific mental illness. To date, there is no “depression” or “anxiety” gene.

Jordan Smoller, MD, ScD, director of the psychiatric and neurodevelopmental genetics unit at Massachusetts General Hospital, said, “Genomic [data] can be evaluated for psychiatric disorders like psychotic illness, where certain copy number variants are well established as having a large effect. But most people don’t have those and for the most part, psychiatric disorders are highly complex, polygenic, and not purely genetic. I wouldn’t rule out the possibility that [polygenic risk scores] could add to what we do in risk prediction, but certainly on their own, they’re not terribly informative at this point.”

Although he is interested in trying to unravel the genetics of mental health, a major part of Smoller’s research focuses on developing clinically useful computational tools for the prediction and early detection of psychiatric disorders and suicidal behavior. By using natural language processing of text and more structured components of the electronic health record (EHR) like diagnostic codes, Smoller’s approach can prospectively sort through patients to find those with a higher risk for suicidal thoughts, which hit an all-time high this past year.

“Clinicians don’t really do much better than chance of knowing who’s really at risk [of suicide], so identifying opportunities for intervention is hampered in that way,” said Smoller. “In the case of suicide risk, what we and others have found is that you can fairly readily train algorithms that perform well and even outperform clinician assessment, and that you can do it in a number of ways. Sometimes you can do it with structured data and EHR diagnostic codes, medications, and procedures that are sort of hard-coded in a sense, or incorporating the narrative text.”

Prospective measurements of brain function

Although using available data to guide physical clinical decision-making, like Smoller does, has been impactful, the elephant in the psychiatric ward is that there need to be significant improvements in the ability to measure and interpret brain function in a clinically meaningful way. Donoghue founded Beacon Bio to address the lack of measurable biological variables for psychiatry and other brain diseases to accelerate clinical trials and enable new treatments for patients.

Early on, while making hardware to measure brain function, Beacon Biosignals acquired a company called Dreem that had developed a headband (called Dreem 3) that was clinically validated for scalable sleep monitoring. In psychiatry, sleep disturbances are not just symptoms, but part of the diagnostic criteria for diseases like major depressive disorder (MDD), bipolar disorder, and post-traumatic stress disorder.

In the past, understanding a patient’s sleep has mostly depended on self-reporting and only recently has it moved to a diagnostic tool called the polysomnogram, which records multiple physiological parameters while a person sleeps and is awake. The Dreem 3—which is more like a fabric-based headset than a TRON-like helmet—has five electroencephalogram (EEG) sensors, a bone conduction speaker for audio output, and a three-dimensional accelerometer to measure movements, head position, and respiratory rate/trace during sleep.

“There are patterns that almost look like language, and our deep learning tools can detect them,” said Donoghue. “There are also features like arousals and sleep spindles, which are these little biomarkers linked to memory and cognition.”

One potential advantage of Beacon Biosignals’ approach with the Dreem 3 is that people can be studied longitudinally. This is critical for addressing challenges in understanding patient responses to medicines over time. The plus side is that patients would not need to come in repeatedly to put on an EEG electrode cap and get data points, but instead can provide continuous data from home by wearing the device over multiple nights.

“By looking at brain function in this way, you can have pharmacodynamic biomarkers,” said Donoghue. “You can see the brain function might be changing immediately, even if potentially clinically you see no change. It might take you four weeks to have a clinical response where you feel better, but maybe you see the semblance of the brain activity changing early on. Maybe it’s helpful to understand why some people are early responders or not or never going to respond. You can imagine clinical paradigms where you start giving this treatment for two nights. You see your brain is not getting engaged with this therapy and maybe you should switch medicines because this is not the right treatment for you.”

The right drug for the right patient

Far before the stage of collecting prospective data on brain function, clinicians and researchers need to stratify and enrich the patient groups most likely to respond to treatment. Back in the 1980s and ‘90s, Prozac, Cymbalta, Paxel, and Wellbutrin were popular brand-name antidepressants, differentiated through a makeshift precision medicine approach that grouped psychiatry patients based on symptoms or lifestyle factors.

“Wellbutrin showed signals of strength in smoking cessation—if you’re a depressed smoker, that’s technically a precision population,” said Rob Houghtaling, senior director of corporate development for Gate Neurosciences. “For Cymbalta, it was depression with some chronic pain. This is the level of precision that psychiatry has had.”

To pair the right patient to the right drug, Gate Neurosciences has developed a precision neuroscience platform for the targeted enrichment of patient subpopulations for their novel NMDA receptor (NMDAR)-targeted therapies that optimize glutamate signaling critical to synaptic function which, when overstimulated, causes long-term depression. Gate Neuroscience’s approach differs from other NMDAR-targeting therapies like ketamine that shut down glutamate signaling. The company has developed NMDAR-positive allosteric modulators that fine-tune glutamate signaling with the goal of achieving rapid and long-lasting treatment outcomes.

For their precision neuroscience platform, Gate Neurosciences incorporates measurements of brain function to identify biomarkers for the company’s clinical pipeline in neuropsychiatry and cognition. In a collaboration with Beacon Biosignals, Gate is using the Dreem 3 headband device and neurobiomarker platform to conduct exploratory EEG and sleep analyses in patients with depression enrolled in Gate’s upcoming Phase II trial of their NMDAR-targeted therapy Zelquistinel, slated to initiate by mid-2024.

Finding new treatments

Amit Etkin, MD, PhD, is trying to raise the bar one step higher—he wants to create completely new drugs that target completely new pathways. Etkin said, “It’s not only in psychiatry that we need better treatments that are personalized, we actually just simply also need new treatments with new mechanisms of action. We’ve been basically mining the same area around monoaminergic antidepressants and mining antipsychotics for a very long time to limited gain.”

Etkin founded Alto Neuroscience, where he is building an industrial process for systematically derisking psychiatric drug development so that every study will actually lower the risk of failure. Alto’s approach has two key components. First, there’s the development of novel drugs that target core brain processes such as emotion, cognition, and sleep, with the goal of improving these functions along with improving symptoms. Second is the development of biomarkers to better identify which patients are more likely to respond to their novel product candidates, which have been identified by evaluating brain function measures like EEG, computerized tests of behavior, and wearables.

The company, which went public in early 2024 with a roughly $129 million IPO, now has five novel drug candidates in Phase II clinical trials for several psychiatric conditions. In July 2024, the company announced that ALTO-100, a novel oral small molecule that has shown evidence of a pro-neurogenesis and neuroplasticity mechanism of action with first-in-class therapeutic potential, recently completed enrollment in Phase IIb for MDD. The predictive biomarker identified, and prospectively replicated, for ALTO-100 is a test of verbal memory, which is considered a direct measure of hippocampal neuroplasticity. Selecting patients based on poor performance on this test provides a clear biological link between the proposed pro-plasticity mechanism of ALTO-100 and this characteristic of responsive patients.

While the field of precision psychiatry is nowhere near linking genes, neurons, brain circuits, and behavior, players in precision psychiatry are on the precipice of understanding how to use existing data, newer tools, and novel ways of thinking to better define specific segments of the population to better understand how it relates to treatment response.

Is this the end of the DSM?

In the past 10 years alone, there has been a complete transformation in the understanding of psychiatric disease states and the pathways involved. With all of the new work done by companies like Beacon, Firefly, Gate, and Alto, it almost seems as if the entire book on precision psychiatry needs to be scrapped and started anew, but that does not seem to be a popular opinion. For Etkin, the reason is that the existing diagnoses and terms have value in grouping people and reaching communities. Smoller’s take is that the diagnoses and terms have become baked into clinical practice, reimbursement, and drug development, making it nearly impossible to shift away. But that could change with what’s being learned from genetics.

“You can identify certain latent underlying genomic factors that really cross these diagnostic boundaries in ways that we don’t think about clinically, and that increasingly is showing us that there’s some fundamental underlying domains or dimensions that present clinically and [that] we categorize in certain ways,” said Smoller. “The same thing is happening to some extent with neuroscience and clinical neuroscience, with the imaging and the kind of transdiagnostic patterns of circuit regulation that we see. So I think the evidence base is growing, but we are not in a position clinically to just replace [the DSM]. I think we can right now raise people’s consciousness and, first of all, [have] humility about the fact that our diagnoses aren’t explanatory, biological, or psychosocial entities, and that they are evolving and they are provisional.”

Putting the stake in the stigma

As precision psychiatry continues to gain momentum, there is the question of whether patients see mental health and psychiatry, which has historically been stigmatized for many reasons, any differently than before.

According to Etkin, “Overwhelmingly, what we’ve seen, and this has been very consistent throughout, is that patients are really bought in. They understand that in fact, they don’t understand why you wouldn’t do a test. They understand that part of having a disease is doing a test, and that creates an objectivity that gets rid of some of the stigma, but even more importantly, it creates a purposefulness in the process of treatment.”

In the near future, maybe in the next five to 10 years, when these precision psychiatry diagnostics and treatments have been approved and commercialized, will people dealing with mental illness suddenly trust their doctors and hop on the train of taking a diagnostic test to get the right psychiatric medication? For the field of psychiatry, that seems like a mighty good problem to have.

Jonathan D. Grinstein, North American editor for Inside Precision Medicine, investigates the most recent research and developments in a wide range of human healthcare topics and emerging trends, such as next-generation diagnostics, cell and gene therapy, genome engineering, and AI/ML for drug discovery. Before IPM Jonathan wrote for publications like Scientific American and Genetic Engineering and Biotechnology News (GEN). Jonathan earned his PhD in biomedical science from the University of California, San Diego, and a BA in neural science from New York University.