Changes in depression symptom network structure following ketamine treatment in treatment-resistant depression.
Joshua Curtiss, Laya Dasari, Julianne Origlio, Catherine Schuessler, Lauren Fisher, Michael D. Kritzer, Minna Behnan, Ava Licht, Nicholas Huempfner, Eric Andrews, Shane Pracar, Sophie Engels, Antonietta Alvarez, Cristina Cusin, Paola Pedrelli
J Affect Disord May 14, 2026 DOI: 10.1016/j.jad.2026.121964 via PubMed
Summary
AI-generated from the abstractPatients with treatment-resistant depression whose symptoms are loosely connected before ketamine therapy are more likely to respond than those with tightly linked symptoms. Among 447 patients receiving intravenous ketamine or intranasal eskatamine, non-responders had denser pre-treatment symptom networks (global strength 4.03) compared with responders (global strength 1.06). After treatment, responders' network strength increased, while non-responders' decreased. Sparse baseline symptom networks may indicate greater capacity for reorganization, making pre-treatment network density a potential marker of ketamine response.
Study at a glance
| Characteristics | Observational cohort Peer reviewed |
|---|---|
| Sample size | 447 |
| Population | Patients with treatment-resistant depression receiving ketamine or esketamine at the MGH Ketamine Clinic |
| Interventions | intravenous ketamine intranasal esketamine |
| Key finding | Pre-treatment symptom network density was significantly higher in non-responders than in responders, and changes in network density after treatment differed between groups. |
Abstract
Major Depressive Disorder (MDD) affects millions globally, with approximately 30% of patients experiencing treatment-resistant depression (TRD). While ketamine has emerged as a rapid-acting intervention, response rates remain variable, underscoring the need to refine precision medicine approaches for ketamine treatment. Network theories of mental health disorders have been promoted as framework for better conceptualizing the structure of symptoms, as well as affording potential insights to bolster personalized treatment approaches. This study sought to leverage a network analytic approach to compare the symptom architecture and density of TRD patients who responded to ketamine treatment versus those who did not. In the current study, 447 patients receiving acute-phase intravenous ketamine or intranasal esketamine at the MGH Ketamine Clinic were included. Gaussian graphical models were estimated using the graphical LASSO method to derive pre- and post-treatment symptom networks (11 nodes) using the QIDS-SR-16. Network density and node centrality were compared between responders and non-responders using permutation-based Network Comparison Tests (NCT). Pre-treatment network density was significantly higher in non-responders (global strength = 4.03) compared to responders (global strength = 1.06; p < 0.01). Following treatment, the responder group showed a significant increase in network strength (to 2.66; p < 0.01), while non-responders showed a significant decrease (to 2.59; p < 0.05). Thus, patients with sparsely connected symptom networks at baseline appear more likely to benefit from ketamine, potentially because their symptoms are more amenable to reorganization. Pre-treatment network density serves as a potential correlate of treatment outcomes of ketamine for TRD.