BACKGROUND: Late-life depression (LLD) is frequently accompanied by cognitive impairment, but short-term treatment-related cognitive change and its predictors remain uncertain. We investigated whether age at first depressive episode, neuropsychiatric symptom phenotype, and baseline peripheral neuroinflammatory biomarkers are associated with cognitive improvement during inpatient treatment for LLD. METHODS: We analysed older inpatients with DSM-IV major depressive disorder from the multi-centre PRODE cohort (n = 136; age ≥ 60). Clinical care was standard, multidisciplinary, and individualised for about eight weeks. Cognition was assessed at admission and discharge using a comprehensive battery. Baseline neuropsychiatric symptoms were measured using the Neuropsychiatric Inventory (NPI), and 12 serum inflammatory markers were collected at admission. RESULTS: Mixed-effects models did not detect overall cognitive improvement across cognitive measures. Late-onset depression (LOD, age 50 years or older) predicted greater improvements in immediate (β = 0.48 95 % CI 0.03, 0.92) and delayed (β = 0.48 95 % CI 0.01, 0.94) recall, which were not significant after correcting for multiple comparisons (ps > 0.08). Latent class analysis (LCA) supported three neuropsychiatric classes. Compared with the Reference class, the Mild class showed larger gains in verbal fluency and delayed recall and greater reduction in Montgomery Aasberg Depression Rating Scale (MADRS) scores, whereas the Severe class did not differ. Lower baseline interleukin-6 (IL-6) and tumour necrosis factor (TNF) - α predicted better recognition memory (β = -0.15, 95 % CI -0.25, - 0.05). CONCLUSIONS: In real-world inpatient care, cognitive improvement in LLD was limited after adjustment. A mild neuropsychiatric profile, and pro-inflammatory biomarkers might be linked to cognitive benefits.
Xue et al. (2026) studied this question.