Metabolic inflammation, brain age and cognitive functioning in short- and long-term clinical weight loss trials

BACKGROUND: Observational studies suggest that metabolic inflammation in obesity can impair brain health, but studies on beneficial effects of weight loss-induced improvements in such markers on brain health and their consequences for clinical outcomes are scarce. METHODS: Consequently, we investigated 53 obese participants in a short-term dietary weight loss trial (up to 4 months, 137 samples; "Muscle Metabolism Study" or "MMS") and 30 in an independent long-term trial (up to 39 months, 100 samples; "Maintain"). For each participant and visit, brain health was characterised in terms of the "brain-predicted age difference" ("brain-PAD"; the difference of the age of a person predicted with machine learning from structural brain MRI minus their chronological age). Increasingly positive brain-PAD scores indicate increasingly poorer brain health. Further, we determined the HOMA index, leptin, fetuin B and CRP levels as markers collectively reflecting low-grade inflammation and impaired metabolic signalling. Finally, we evaluated the relevance of these parameters for brain-PAD and the association of brain-PAD alterations for cognition, which was measured in the MMS with neuropsychological tests. FINDINGS: Weight loss led to improved brain-PAD scores (MMS: t = -2.02, p = 0.023, effect size partial η(2) (η(2)(p)) = 0.03; Maintain: t = -7.37, p = 4.2·10(-11), η(2)(p) = 0.38). According to a False Discovery Rate (FDR) method-corrected threshold (α(FDR) = 0.05), HOMA index (MMS: t = 2.28, p(FDR) = 0.024, η(2)(p) = 0.04; Maintain: t = 2.33, p(FDR) = 0.023, η(2)(p) = 0.08), and leptin (MMS: t = 4.43, p(FDR) = 4.3·10(-5), η(2)(p) = 0.14; Maintain: t = 1.91, p(FDR) = 0.041, η(2)(p) = 0.06), showed significant positive links to brain-PAD in both trials, fetuin B did so in Maintain (t = 2.57, p(FDR) = 0.023, η(2)(p) = 0.11). Brain-PAD variations were associated with a neuropsychological test of psychomotor speed and visual attention (t = 2.32, p(FDR) = 0.022, η(2)(p) = 0.05). Application of explainable artificial intelligence methods showed that this link was parallelled by widespread brain age-related tissue alterations in white and grey matter involved in these functions. INTERPRETATION: Analyses of two independent weight loss trials suggest that weight loss-induced improvements in metabolic-inflammatory markers have beneficial effects on brain-PAD and the latter were associated with enhancements in cognitive functioning, underscoring the potential clinical relevance of metabolic brain age regulation.