Psychiatric drug discovery is gaining renewed pharmaceutical and investor interest after years of central nervous system R&D retreat.
High placebo response, subjective endpoints and heterogeneous patient populations have made psychiatric clinical trials difficult to interpret and de-risk.
Biomarkers, functional imaging and patient stratification may help improve signal detection in neuropsychiatric drug development.
Recent approvals, such as Cobenfy and Zurzuvae ,show that differentiated psychiatric medicines can still reach regulatory approval.
Major acquisitions and financings involving Karuna, Cerevel, Intra-Cellular Therapies, Leal Therapeutics and Draig Therapeutics suggest selective capital return to CNS and neuropsychiatric pipelines.
A Large Unmet Need, but a Difficult Field to Drug
Psychiatric drug discovery has never lacked urgency. Mental health disorders represent a vast global burden, with the World Health Organization (WHO) estimating that 970 million people were living with a mental disorder in 2019. Depression alone affected an estimated 280 million people that year, while schizophrenia affects approximately 23 million people worldwide.1 These figures make clear that the retreat from psychiatric R&D was not driven by a lack of patients, unmet need, or public-health relevance.
The harder problem has been translation. Psychiatric disorders are defined largely by clinical symptoms and behavioral patterns, but those diagnostic categories do not always map cleanly to discrete biological mechanisms. That creates a difficult starting point for drug discovery: developers must often pursue targets in diseases where the relationship between underlying biology, patient presentation, clinical course, and treatment response remains incomplete. Recent neuropsychiatric drug development literature identifies this weak connection between biology and disease as one of the central challenges facing the field.2
That disconnect has shaped the risk profile of psychiatric drug development. In many therapeutic areas, stronger mechanistic anchors can help developers connect target biology to patient selection, pharmacodynamic evidence, clinical endpoints, and regulatory strategy. In psychiatry, those links have often been harder to establish. A compound may have a plausible mechanism, but the trial population may include biologically heterogeneous patients; a clinical endpoint may capture meaningful change, but only indirectly; and an efficacy signal may be difficult to separate from variability in disease course, care context, or placebo response. The result is a field with enormous need but a development pathway that has historically been difficult to de-risk.
That paradox sets the stage for the current moment. Psychiatry remains one of the areas where better medicines could have an enormous impact, but rebuilding the pipeline requires more than renewed enthusiasm. It requires development strategies that can connect mechanisms to patients, patients to endpoints and endpoints to signals that are strong enough to justify continued investment.
Why Pharma Pulled Back from CNS and Psychiatric R&D
The scale of unmet need did not prevent companies from backing away from central nervous system (CNS) and psychiatric research. By the early 2010s, the industry’s retreat had become visible enough to prompt broad concern across the field. Several high-profile pharmaceutical companies shut down major CNS research activities, including Novartis, GlaxoSmithKline, AstraZeneca, Pfizer, Merck, and Sanofi. At the time, the withdrawal of pharmaceutical companies from neuroscience research was viewed as a serious concern linked to the high-risk nature and limited success of new psychotherapeutic drug development.3
That pullback reflected a hard commercial and scientific calculation. CNS programs often required long development timelines, complex clinical trials and substantial investment before companies could know whether a mechanism would translate into meaningful clinical benefit. Contemporary reporting described brain-disease drug development as more difficult, often more time-consuming and expensive and therefore a weak investment area, leading several large pharmaceutical companies to selectively downsize neuroscience research divisions.4
For psychiatric drug development specifically, this created a difficult feedback loop. The field clearly needed better medicines, but the development model often struggled to produce predictable signals of success. When target biology was uncertain, patient populations were heterogeneous, and clinical outcomes were difficult to measure cleanly, even well-resourced programs could become hard to justify. The result was not a disappearance of interest in mental health, but a loss of confidence in the ability to turn psychiatric biology into de-risked, approvable, and commercially viable therapies.
Today’s renewed interest is emerging against the memory of that retreat. Companies are not returning to psychiatry simply because the need is large; that need was always present. They are returning where there is a stronger case that the scientific and clinical-development risks can be better understood, better managed and, in some cases, reduced enough to make investment plausible again.
The Clinical Trial Problem: Placebo Response, Subjective Endpoints, and Signal Detection
The risks that pushed companies away from psychiatric R&D were not limited to target discovery or early biology. They also appeared in the clinic, where central nervous system programs have long struggled to produce clear, reproducible evidence of efficacy. CNS drug development has been hampered by repeated negative and failed placebo-controlled trials, and perceived trial risk contributed to the broader withdrawal of many large pharmaceutical companies from the field.5
A central issue is signal detection. In psychiatry, a negative trial can mean that a molecule is ineffective, but it can also mean that the trial was not able to distinguish a true drug effect from noise. Patient populations may be broad or biologically heterogeneous, and study enrollment may include participants whose symptoms look similar clinically but reflect different underlying mechanisms or trajectories. Misclassification of study subjects has been identified as one of the major contributors to CNS trial failure, alongside excessive placebo response.5
Placebo response creates a particularly difficult problem because it can shrink the apparent difference between treatment and control arms. High and growing placebo response rates in CNS trials, including trials in depression and schizophrenia, have posed a major challenge for drug development.6 In this setting, a compound may need to demonstrate efficacy against a moving and sometimes unexpectedly strong comparator, making it harder to interpret whether a failed study reflects weak pharmacology, trial design limitations or the difficulty of measuring psychiatric improvement in a controlled trial environment.
Measurement adds another layer of complexity. CNS trials often rely on clinician-rated assessments, which can be affected by variability and bias. These measures can capture clinically meaningful change, but they are not the same as a direct biological readout. When endpoints depend on interviews, rating scales, and symptom interpretation, trials become more vulnerable to site-to-site differences, rater variability, and expectation effects. That does not make the endpoints irrelevant; it makes trial execution and patient selection especially important.
The cumulative effect has been a development environment with unusually high attrition. A 2021 study found that psychiatric medications had a 6.3% ultimate likelihood of approval, with 23.7% phase II success and 55.7% phase III success.6 Those figures should not be treated as a fixed benchmark for every psychiatric program today, but they illustrate why the field came to be viewed as difficult to finance and difficult to de-risk. For companies deciding where to place capital, psychiatric trials often appeared to combine high unmet need with uncertain biology, subjective endpoints, and a persistent risk that even a promising mechanism might fail to separate from placebo.
The Biology Problem: Diagnoses Do Not Always Map Cleanly to Mechanisms
The clinical trial challenges in psychiatric drug development are closely tied to a deeper biology problem: diagnostic categories do not always provide a precise map for drug discovery. Conditions like depression, schizophrenia, and other psychiatric disorders are defined primarily through clusters of symptoms, functional impairment, and clinical history. Those categories are essential for diagnosis and care, but they can be blunt tools for therapeutic development when the goal is to connect a molecular target to a specific disease-driving mechanism.
This weak connection between underlying biology and disease is one of the central difficulties in neuropsychiatric drug development. In practice, a single diagnosis may include patients with different biological drivers, different illness trajectories, and different likelihoods of responding to a given mechanism. At the same time, the same biological dysfunction or symptom domain may cut across multiple diagnoses. That makes it difficult to know whether a therapy has failed because the mechanism was wrong, the dose was wrong, the endpoint was insensitive, or the study population was too biologically mixed.
This creates a challenge for traditional indication-based development. A program built around a broad diagnostic label may enroll a clinically appropriate population but still dilute the very signal it is trying to detect. If only a subset of patients has the biology a drug is designed to modulate, the effect in that subgroup may be obscured when the full trial population is analyzed together. That risk helps explain why patient selection and translational evidence have become so important to rebuilding confidence in the field.
For that reason, newer thinking in neuropsychiatric development increasingly emphasizes the need to focus on discrete biological dysfunctions and symptom domains rather than diagnoses alone.7 This does not mean diagnostic categories should be discarded. Rather, it suggests that psychiatric drug development may need a more layered model, one that starts with clinical diagnosis but adds biological, functional, and symptom-domain information to better identify which patients are most likely to benefit from a given therapy.
That shift is central to the rebuilding of the psychiatric pipeline. If companies can connect mechanisms more directly to the patients and symptoms those mechanisms are most likely to affect, the field may become less dependent on broad, heterogeneous trial populations and more capable of generating interpretable clinical signals.
What Is Changing: Biomarkers, Stratification, and Better Trial Design
If the earlier era of psychiatric R&D was defined by uncertainty, the current rebuilding effort is increasingly focused on making that uncertainty more manageable. The field has not solved the core challenges of psychiatric drug development. Diagnosis remains imperfect as a proxy for biology, placebo response remains a major concern, and many clinical endpoints still depend on symptom-based assessments. What has changed is that developers have a clearer view of where trials break down and a growing set of tools for making studies more biologically and clinically informative.
One important shift is the use of objective measures to strengthen confidence earlier in development. Functional imaging has been proposed as a tool that can provide information across clinical phases, including data related to brain function, CNS penetration, dosing, target engagement, efficacy, side effects, responder status, and differences between placebo and drug response.5 In a field where clinical symptoms may not map neatly onto biological mechanisms, those types of measures can help developers ask more precise questions: Does the drug reach the brain? Does it engage the intended target or pathway? Is the biological effect consistent with the proposed mechanism? Are there early signals that can inform dose selection or patient selection before a large efficacy trial?
Biomarker-driven stratification offers another path toward more interpretable studies. Biomarker-based patient stratification in phase II and III studies offers a way to increase sensitivity and statistical power while reducing costs. The rationale is straightforward: if a therapy is designed to act on a particular biological dysfunction, then development should, where possible, enrich for patients most likely to have that dysfunction. That approach may help avoid a common problem in psychiatric trials, where a potential treatment effect in one subgroup is diluted across a broader and more heterogeneous population.
Trial design is also becoming more deliberate about placebo response. In psychiatric indications, where placebo response can obscure efficacy, the way patients are educated, enrolled, assessed and followed can influence the ability to detect a real treatment effect.
These changes matter because they shift the field away from simply testing broad diagnostic populations and hoping for a clean clinical signal. A more modern psychiatric development program may combine symptom-based diagnosis with biological evidence, functional measures, stratification strategies, and trial procedures designed to reduce noise. None of these tools guarantees success. Together, however, they can make psychiatric trials more informative and give companies a stronger basis for deciding whether to advance, modify or stop a program.
Recent Approvals as Confidence-Building Events
Recent psychiatric drug approvals have helped change the tone around the field. They do not erase the scientific and clinical-development barriers that pushed many companies away from CNS R&D, but they do show that differentiated psychiatric medicines can still move through development and reach regulators successfully. That matters for a field where confidence was weakened by repeated trial failures, difficult signal detection and uncertainty about whether new mechanisms could translate into approvable therapies.
Cobenfy™ is one of the clearest examples. The U.S. Food and Drug Administration (FDA) approved Cobenfy in September 2024 for the treatment of schizophrenia in adults. The approval drew attention not only because of the indication, but because FDA described Cobenfy as the first antipsychotic approved for schizophrenia that targets cholinergic receptors rather than dopamine receptors.8 For a therapeutic area long associated with familiar pharmacologic approaches, that mechanism helped position the approval as evidence that new biological strategies can still break through in psychiatry.
Zurzuvae® offers another confidence-building example. The FDA approved Zurzuvae in August 2023 as the first oral medication indicated to treat postpartum depression in adults.9 Like Cobenfy, the approval is best understood as a proof of principle rather than a general solution to psychiatric drug development. It shows that clinically meaningful psychiatric innovation can reach approval, but it does not remove the need for better trial design, more precise patient selection, or stronger translational tools across the broader field.
The significance of these approvals is partly scientific and partly psychological for the industry. There appears to be growing optimism in the field because of recent approvals and renewed interest from pharmaceutical companies and investors.2 That optimism is emerging after years in which the field’s high failure rates made many companies cautious. Recent approvals provide a counterweight to that history by showing that psychiatric development can still produce regulatory successes when a program has a clear rationale, a defined clinical need and evidence strong enough to support approval.
The most useful way to frame this moment is not as a sudden reversal of all the field’s problems. Psychiatry remains difficult to drug, and the barriers discussed earlier still shape development decisions. Approvals like Cobenfy and Zurzuvae help make the field investable again by demonstrating that innovation is possible, that regulators will recognize differentiated approaches, and that psychiatric indications can still support meaningful product development when the science and clinical strategy align.
M&A and Financing Signals: Capital Is Returning, but Selectively
Recent dealmaking suggests that the renewed interest in psychiatric and broader CNS drug development is not limited to the literature. Capital is moving back into the field, but selectively. The strongest signals are forming around assets with clinical validation, differentiated mechanisms, or pipelines that give larger companies a clearer strategic rationale for re-entering neuroscience.
Bristol Myers Squibb’s acquisition of Karuna Therapeutics became one of the most visible examples of that shift. The company agreed to acquire Karuna for $14.0 billion and completed the acquisition in March 2024. In announcing the transaction, Bristol Myers Squibb emphasized that Karuna would add KarXT, or xanomeline-trospium, which ultimately became marketed as Cobenfy.10 In the context of a field long constrained by uncertainty around mechanism, trial outcomes and commercial confidence, that type of asset offered a clearer reason for a major pharmaceutical company to invest.
AbbVie’s acquisition of Cerevel Therapeutics provided another major signal. AbbVie agreed to acquire Cerevel in a transaction valued at approximately $8.7 billion and completed the acquisition in August 2024.11 Together, the Karuna and Cerevel deals suggested that large pharmaceutical companies were again willing to pursue neuroscience and psychiatric pipelines where they saw differentiated biology and a credible path to value creation.
That pattern continued into 2025. Johnson & Johnson announced in January 2025 that it would acquire Intra-Cellular Therapies for approximately $14.6 billion, with the transaction including CAPLYTA and a clinical-stage pipeline spanning generalized anxiety disorder and Alzheimer’s disease-related psychosis and agitation. The company completed the acquisition in April 2025.12 Like the Karuna and Cerevel transactions, the deal reinforced the idea that neuroscience assets are again attracting major strategic attention when they offer approved products, late-stage opportunities or pipeline depth in areas of high unmet need.
The return of capital is also visible outside large pharmaceutical acquisitions. Leal Therapeutics announced a $45 million financing in October 2024 to advance CNS programs, including a small-molecule program for schizophrenia.13 Draig Therapeutics received $140 million in funding to advance therapies for major neuropsychiatric disorders, including a lead candidate for major depressive disorder.14 These financings point to renewed investor willingness to support earlier-stage neuropsychiatric companies, particularly when programs are framed around defined mechanisms and serious disorders with limited treatment options.
The broader neuroscience capital environment also appears more active. DealForma’s 2025 neurology review reported stronger merger and acquisition growth, steady venture funding and improved initial public offering activity in neurology, although that analysis should be interpreted as a broader neuroscience signal rather than a psychiatry-specific measure.15 For psychiatric drug discovery, the more precise conclusion is that capital is returning where investors and strategic buyers see ways to reduce uncertainty: approved or late-stage assets, novel mechanisms, differentiated clinical profiles and pipelines that fit larger neuroscience strategies.
That selectivity is important. The field is not returning to an era of broad, unconstrained enthusiasm for any psychiatric target. The deals and financings instead suggest a more disciplined rebuilding of the pipeline, with capital flowing toward programs that appear better positioned to overcome the development barriers that drove the earlier retreat.
What the Next Generation of Psychiatric Drug Development Must Get Right
The return of capital does not mean psychiatric drug development can return to older assumptions. The next generation of programs will need to be more deliberate than the one that preceded the industry’s retreat. Large, diagnosis-defined trials may still be necessary for approval, but the path to those trials will need stronger biological rationale, clearer translational evidence, and better ways to identify the patients most likely to respond.
The first requirement is a tighter link between mechanism and disease. A rebuilt pipeline cannot rely only on the presence of a large unmet need or a plausible target. It must show why a mechanism matters for a defined patient population, how that mechanism can be measured, and how modulation of that mechanism should translate into clinical improvement.
That need is reshaping the role of biomarkers. Biomarker strategies may improve development by focusing on discrete biological dysfunctions or symptom domains rather than diagnoses alone. This is especially important in psychiatry, where the same diagnosis can encompass patients with different biological profiles and where the same symptom domain may appear across different disorders. A biomarker-informed strategy can help developers move beyond broad clinical labels and ask more specific questions about who should be enrolled, what biological effect should be expected, and what early evidence would justify continued investment.
Patient stratification is likely to be central to that effort. Biomarker-based stratification in phase II and III studies may increase sensitivity and statistical power while reducing costs. For psychiatric trials, this could be particularly valuable because heterogeneous enrollment can dilute a true treatment effect. If developers can identify subgroups more likely to respond to a given mechanism, trials may become smaller, cleaner and more informative, even if the overall field remains complex.
Trial execution will also need to reflect the lessons of past failures. Psychiatric trials must account for placebo response, subjective endpoints, and patient heterogeneity from the beginning, not treat them as problems to explain after a negative result. That means placing greater emphasis on rater training, endpoint selection, enrollment criteria, placebo-response mitigation, and objective measures that can support interpretation of clinical outcomes. The goal is not to eliminate uncertainty but to design studies that are less vulnerable to avoidable noise.
The rebuilt psychiatric pipeline will therefore need to be mechanism-led, biomarker-informed, patient-stratified, and trial-design-conscious. Those principles will not guarantee success, but they address the specific weaknesses that made the field so difficult to de-risk in the past. For companies returning to psychiatry, the central question is no longer whether the unmet need justifies investment. It is whether each program can generate evidence strong enough to make that investment rational.
The same issues that drove the earlier retreat will continue to shape which programs attract capital and which fail to advance. Psychiatric drug discovery is becoming investable again where developers can build stronger mechanistic rationales, identify better-defined patient populations, generate evidence of target engagement and design trials capable of producing interpretable clinical signals. The opportunity is not simply to restart the old pipeline but to rebuild it around the lessons that past failures made impossible to ignore.
References
1. “Mental Health.” World Health Organization. Accessed 11 Mar. 2026.
