Subtyping depression using brain-gastric electrophysiology for early prediction of antidepressant response: a...

Depression affects approximately 5% of adults worldwide, with India reporting a prevalence of 4.5%. Oral medication is a common treatment, but over 50% of patients fail to respond to first-line antidepressants and often require medication adjustments or augmentation. This highlights the urgent need for predictive models that can guide personalized treatment strategies more quickly. Our study aims to achieve three key objectives: first, to assess the predictive ability of previously identified biomarkers such as frontal theta power and alpha asymmetry in explaining the response to interventions within our sample; second, to evaluate the utility of a holistic research approach focusing on gut-brain interactins, in predicting responses; and third, to identify reliable early biomarkers that can predict responses across various phenotypic subtypes characterized by dominant clinical presentations. Across two sites (site 1 + site 2), a total of 206 (161 + 45) participants, including 144 (99 + 45) treatment-naive patients, were enrolled in our study, which spanned three visits. We aimed to predict antidepressant outcomes at the third visit (4-6 weeks) using data collected from visits one (baseline) and two (7-10 days), and the data from site 2 were solely used for testing the predictive utility. Our predictive models, which incorporated electrophysiological data from both the brain and gut along with clinical information, achieved cross-validation (holdout test) performance of 78% (80%) specificity and 84% (71.43%) sensitivity in identifying nonresponders to antidepressant treatment administered as per Clinical Practice Guidelines of India. We found that certain electrophysiological features were strongly predictive of treatment outcomes for specific depression subtypes. For example, increased excitation-inhibition ratios in the frontocentral brain regions were predictive of patients with dominant anxiety and sleep symptoms. Similarly, decreased tachygastric coupling with the sensorimotor brain region predicted treatment nonresponse in patients with high levels of negative self-thoughts. Increased connectivity in the right frontocentral region was associated with better outcomes in patients with significant appetite issues. Additionally, higher frontocentral theta power and beta asymmetry were predictive of responses in patients with a composite set of symptoms. Our findings suggest that combining brain and gut electrophysiological markers with clinical phenotyping offers a promising, scalable approach to personalize depression treatment. This approach could guide clinicians in developing more effective and tailored medication strategies, ultimately improving patient outcomes.

Clinical identifier Trial Registry of India https://ctri.nic.in/Clinicaltrials, CTRI/2025/04/084404.