Optimizing Study Design for Clinically Useful Peripheral Molecular Biomarker Discovery in Psychiatry
Chloe X. Yap, Anjali K. Henders, Naomi R. Wray, Rachel UpthegroveImportance
Psychiatric diagnosis, prognosis, and management currently rely on subjective assessments. There is hope that peripheral (eg, blood) biomarkers or combinations thereof could transform clinical care by bringing objectivity and patient stratification while being accessible and scalable. The time is ripe to find molecular biomarkers: omics technologies can now profile hundreds to millions of molecules across thousands of individuals, while funders recognize the clinical need and have responded with ambition. However, this enthusiasm will likely be finite, and there is no guarantee that a biomarker will be found.
Observations
It is essential to design and perform studies that are most likely to deliver robust peripheral biomarkers with a clear path to translation. This narrative review provides recommendations to optimize peripheral psychiatric biomarker study design. The review first introduces conceptual grounding illustrating how biomarker study design must triangulate clinical utility, technological fit, and biological plausibility. Biomarker studies should optimize for signal-to-noise ratio via careful phenotype selection and minimization of biological and technical noise. However, feasibility issues—in collecting sufficient data on enough representative participants for an adequately powered and clinically translatable analysis—constrain study design.
Conclusions and Relevance
Study design is the single most influential and controllable factor for future psychiatric biomarker discovery. This is because many years elapse between study design and data analysis (particularly for biobanked samples held for future use) and because no technological or analytical advances can answer questions that the dataset was not designed for, or overcome unmeasured confounding. Moreover, the direction of biomarker research is constrained by what is possible to design. Many clinical questions and contexts need biomarkers, but realistically, some are better primed for biomarker discovery than others. In this generation of peripheral psychiatric biomarker studies, designs most likely to demonstrate proof-of-principle should be prioritized to ensure the long-term sustainability of the field.