Association between depression subtypes and response to repeated-dose intravenous ketamine.Objective: About half or more of treatment-resistant depressed patients do not respond to ketamine, and few clinical predictors to gauge the most likely antidepressant response have been proposed. We explored whether depression subtypes are associated with response to ketamine. Method: Ninety-seven participants with depression were administered six repeated-dose intravenous ketamine and assessed for depression (Montgomery-Asberg Depression Rating Scale, MADRS), anxiety (Hamilton Anxiety Rating Scale, HAMA), and suicidal ideation (Beck Scale for Suicidal Ideation, SSI) at baseline, 24 h after each infusion, and 2 weeks after the whole treatment. Participants were classified by melancholic/ anxious subtype. Individuals who met criteria for neither or both subtypes were classified separately, resulting in four mutually exclusive groups. Results: Patients with melancholic or melancholic-anxious features were less likely to respond (e.g., day 13, melancholic-anxious vs. anxious, OR 0.138, 95% CI 0.032-0.584, P = 0.007) or remit (e.g., day 26, melancholic vs. no subtype, OR 0.182, 95% CI 0.035-0.960, P = 0.045) and took longer to achieve response/remission than those with anxious or no subtype features. Faster HAMA score reductions were observed in patients with anxious or melancholic-anxious features, and faster SSI score reductions were observed among those with melancholic-anxious features. Conclusion: Our study shows promising results for ketamine as a novel antidepressant preferentially for the treatment of non-melancholic or anxious depression. Significant outcomes• Compared to those with pure anxious or no subtype features, patients with pure melancholic or melancholic-anxious features were less likely to respond or remit.• Patients with melancholic or melancholic-anxious depression took longer to achieve response or remission and generally improved less in depression, anxiety, and suicidal ideation than those with pure anxious features or no subtype features.• All subtype groups exhibited significant symptom reductions throughout the treatment, while faster score reductions on HAMA were observed in patients with anxious or melancholic-anxious features, and faster score reductions on SSI-I and SSI-total were shown in patients with melancholic-anxious features. Limitations• Only two depression subtypes (i.e., melancholic and anxious) were classified in the present study drawing into question the confounding effect of other subtypes.• All patients included were treatment-resistant or with suicidality, which limits the generalizability of our results to patients with depression as a whole.• Although patients continued medications that they were receiving at screening at the same stable dosages throughout the study, we cannot rule out the influence of these medications on depressive symptoms.