Psychiatric precision medicine has been a promising area of research for nearly two decades, but only recently has data emerged that can influence real clinical trial design. In 2025, several developments suggest that stratification based on genetics, pharmacokinetics, and mechanistic biology is becoming more actionable. New genetic responder signals, validated EEG-based subtypes, and the first biomarker-enriched depression trial have pushed the field past its long-standing theoretical phase.
This blog summarizes the major drivers behind this shift, including newly reported subgroup signals, real-world data infrastructure, and validated pharmacogenomic evidence.
Compared to other therapeutic areas in precision medicine, psychiatry has experienced a slower pace of development. Unlike monogenic or pathway-defined diseases where gene therapy is effective, psychiatric disorders involve hundreds of genetic variants with very small effects, none of which define a clear molecular target for intervention. Because the relevant dysfunctions are network-level, not gene-level, psychiatric disorders have not followed the same gene therapy trajectory as rare neurological or metabolic diseases.
These realities explain why precision psychiatry has evolved through a different set of tools. Instead of correcting single genes, progress has focused on identifying subgroups defined by clinical patterns, pharmacokinetic variation, and emerging biomarkers. While decades of genetic studies have revealed molecular signatures underpinning disease risk, these have been difficult to translate into actionable clinical tools.
Psychiatric disorders encompass multiple subphenotypes within a single diagnosis, making it challenging to identify differences in treatment response across subgroups. Heterogeneity in symptom presentation and illness trajectory reduces the power to detect consistent biological patterns.
Large-scale genome-wide association studies (GWAS) have identified hundreds of common variants associated with major psychiatric disorders. However, these variants are rarely linked to biological targets of current treatment modalities. As a result, these variants are unlikely to serve as predictors of treatment response.
Most variants associated with psychiatric disorders carry small effect sizes, requiring extremely large datasets to detect reliable phenotype–genotype relationships. Trial cohorts have historically been too small to support biomarker discovery.
Placebo response rates are known to be high in psychiatric disorders, particularly in depression and anxiety. This masks modest treatment effects and reduces the ability to detect subgroup-specific responses. The absence of validated biological markers further complicates patient stratification efforts.
Several foundational advances over the past decade created the necessary conditions for meaningful precision strategies to develop. These advances span pharmacogenomics and computational biomarker discovery.
Several studies have confirmed that pharmacokinetic variability is a source of heterogeneity in depression treatment. For instance, escitalopram exposure differs substantially across CYP2C19 genotypes, with poor metabolizers exhibiting much higher serum concentrations and increased rates of treatment discontinuation or switching, while ultrarapid metabolizers show lower concentrations and higher treatment failure.
Although pharmacogenomics does not identify drug-response biology, these consistent pharmacokinetic subgroups demonstrated that stratification is possible and clinically relevant in depression.
Advances in computational biology are facilitating the discovery of novel biomarkers in psychiatry. Recently, machine learning applied to baseline EEG data revealed subgroups with distinct antidepressant trajectories: drug responders, non-responders, placebo responders, and individuals who worsened on sertraline. These clusters were stable in cross-validation and reproducible across trial sites.
Together, pharmacogenomics, real-world data, and AI-based neural biomarkers have established the evidence base needed for biomarker-guided clinical trials to become more feasible.
The advances described above created the foundation, but 2025 brought the first clear clinical demonstration that biologically defined psychiatric subgroups can drive decision-making in drug development.
One example comes from HMNC Brain Health, a precision psychiatry biotech based in Germany, which recently reported that its Phase IIb trial in major depressive disorder (MDD) identified a predefined genetic subgroup that responded more strongly to treatment. The program’s investigational therapy, BH-200, targets the CRHR2 receptor involved in stress-response regulation, and the genetic signal appeared to map onto this mechanism. Based on this finding, the company is preparing a Phase III program enriched for individuals who carry that genetic signature.
This milestone is notable for several reasons:
This result can be viewed as the first clinical signal emerging from years of progress across genomics and biomarker development.
The convergence of these signs directly affects how psychiatric trials can be designed going forward.
Biologically enriched recruitment strategies: Sponsors can incorporate genetic, pharmacokinetic, electrophysiological, or digital markers to recruit participants who share a more homogeneous underlying biology. This reduces noise in efficacy outcomes and increases the probability of detecting true drug effects.
Adaptive and stratified trial designs: Modern trials can include predefined subgroup expansions, response-adaptive randomization, or mid-study enrichment based on early biomarker performance. These approaches align with emerging regulatory acceptance of precision strategies in CNS trials.
Dose optimization based on metabolic genotype: Pharmacokinetic-related genotypes offer a straightforward path for exposure-based stratification. Trial protocols can use genotype information to optimize dosing, manage adverse events, and reduce attrition, improving overall study quality.
Multimodal biomarker frameworks: Combining genotypes, clinical phenotypes, EEG signatures, and real-world outcomes may offer the clearest picture of responder subgroups. Real-world data can help validate these multimodal signatures by showing whether the same subgroups appear across much larger and more diverse patient populations, based on real prescribing patterns and longitudinal outcomes.
Psychiatric precision medicine is entering a new phase. For decades, the field struggled with polygenic architectures, high heterogeneity, and a lack of targetable molecular mechanisms. Foundational advances in pharmacogenomics and AI-based biomarker discovery have now created the necessary scaffolding for clinically meaningful stratification.
The HMNC BH-200 subgroup signal is the strongest indication to date that biologically anchored responder groups can be identified in depression and used to shape Phase III development. As these approaches expand, psychiatric trials may increasingly mirror precision medicine frameworks used in other therapeutic areas.
To learn how Sano accelerates biomarker-driven and patient-centered neurology trials, read Why Sano for Neurology Trials.