A predictive modelling pipeline for deriving treatment response-sensitive personalised neural circuits for anxiety and depression

Anxiety and depression are the most common psychiatric conditions globally and considered leading causes of disability by the World Health Organization (WHO), with combined lifetime prevalence estimates of around 30% for anxiety and 20% for depressive disorders (Kessler et al., 2005). Current first-line treatments such as pharmacotherapy, which involves antidepressant medications such as escitalopram, and psychotherapy, most commonly cognitive behavioural therapy (CBT) are available, but remission rates around 50% leave substantial room for improvement (Fava, 2003; Springer et al., 2018). Despite a strong evidence base for their efficacy, the allocation to these treatments relies in large part on a trial-and-error approach, due to the absence of evidence to guide clinicians in predicting a priori which treatment a patient is likely to benefit from. This highlights the importance of predictive modelling in precision psychiatry to develop prescriptive biomarkers, which characterise neurobiological dysfunctions for an individual patient at baseline and predict on that basis which treatment they are likely to respond to. To address this gap, this study introduces a novel predictive modelling pipeline that integrates theory-driven neural circuits with data-driven regularization to derive personalized treatment response-sensitive circuit scores for predicting anxiety and depression response to psychotherapeutic or pharmacological intervention. Using baseline functional magnetic resonance imaging (fMRI) data from 141 anxious-depressed patients and a reference sample of 97 healthy controls, twenty-two features of dysfunctional task-based activation, task-based connectivity, and task-free connectivity were extracted across five theoretically defined functional circuits: negative affect, positive affect, default mode, salience, and attention. To regularize these features, ridge regression models were trained to predict post-intervention clinical outcomes based on depression (PHQ-9) and anxiety (GAD-7) scores across an escitalopram, placebo, cognitive behavioural therapy (CBT), and waitlist intervention. The twenty-two features were aggregated into the five circuits based on their prediction strength for each clinical outcome. These resulting personalized circuit scores were then tested as moderators of treatment response in two randomized controlled trials comparing cognitive behavioural therapy (CBT; n=63) versus a waitlist control (n=35), and escitalopram (n=21) versus a placebo group (n=22). This revealed that decreased default mode circuit engagement at baseline predicted greater depression improvements following CBT versus waitlist (β=1.209, p=0.010). Conversely, increased salience circuit engagement predicted greater anxiety reduction following escitalopram versus placebo (β=-2.314, p=0.028). No other circuits significantly moderated treatment response for their respective outcomes. This study provides preliminary evidence that personalized neural circuit biomarkers can prospectively predict differential treatment response in anxiety and depression. The novel analytical pipeline balances interpretability with predictive performance and tests for treatment-specific effects across both active and control interventions for pharmacotherapy and psychotherapy. While requiring validation in larger independent samples, these findings represent a promising step toward developing precision psychiatry approaches that offer prescriptive biomarkers to guide individual-level treatment selection.

References

Kessler, R. C., Berglund, P., Demler, O., Jin, R., Merikangas, K. R., & Walters, E. E. (2005). Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the National Comorbidity Survey Replication. Archives of General Psychiatry, 62(6), 593–602. https://doi.org/10.1001/archpsyc.62.6.593 Fava, M. (2003). Diagnosis and definition of treatment-resistant depression. Biological Psychiatry, 53(8), 649–659. https://doi.org/10.1016/S0006-3223(03)00231-2 Springer, K. S., Levy, H. C., & Tolin, D. F. (2018). Remission in CBT for adult anxiety disorders: A meta-analysis. Clinical Psychology Review, 61, 1–8. https://doi.org/10.1016/j.cpr.2018.03.002