Journal: Alzheimer's Research & Therapy

Article Title: Functional network phenotypes of mild behavioural impairment: cognitive effects moderated by amyloid

doi: 10.1186/s13195-026-01980-2

Figure Lengend Snippet: Study schematic. ( A ) Univariate associations between FC matrices and MBI diagnosis, MBI-C total score or MBI-C subdomain score were examined using separate linear regression models, with age, sex, years of education, diagnosis and total intracranial volumes included as nuisance covariates. ( B ) In parallel, partial least squares correlation was conducted to examine multivariate associations between residuals of FC matrices and MBI-C subdomain scores after regressing out age, sex, years of education, diagnosis and total intracranial volumes. This approach gives rise to a set of latent variables, which are linear weighted combinations of the original variables (i.e., FC and MBI-C score loadings) that have maximal covariance with each other. Individual connectome scores and MBI-C scores were then obtained by back projecting the FC and MBI-C score loadings to their original residual values. Connectome scores describe the extent to which each participant expresses the FC pattern maximally associated with the MBI-C scores, with higher connectome scores indicating greater MBI-related functional network disruptions. ( C ) Subsequently, we examined whether connectome score or MBI-C total score interacted with global amyloid SUVR and temporal meta-ROI tau SUVR to influence baseline and rate of change in global cognition and functional performance using linear regression models. MBI-C = Mild Behavioural Impairment Checklist; MBI-C = Mild Behavioural Impairment Checklist; SUVR = standardized uptake value ratio; ROI = region-of-interest; FC = functional connectivity; AD = Alzheimer’s disease; CDR = Clinical Dementia Rating; MoCA = Montreal Cognitive Assessment

Article Snippet: Behaviour partial least squares correlation was then performed on the standardized FC and MBI-C subdomain score residuals using the PLS toolbox [ ] in MATLAB.

Techniques: Biomarker Discovery, Functional Assay

Journal: Alzheimer's Research & Therapy

Article Title: Functional network phenotypes of mild behavioural impairment: cognitive effects moderated by amyloid

doi: 10.1186/s13195-026-01980-2

Figure Lengend Snippet: The presence and severity of MBI are associated with whole-brain FC dysfunctions. ( A ) FC matrix (left) displays significant bootstrap ratios (> 2) of functional connections corresponding to this latent variable, while bar chart (right) displays the mean correlation values between connectome scores of this latent variable and each of the MBI-C subdomain scores (error bars denote 95% bootstrapped confidence intervals). Partial least squares correlation analysis identified one significant latent variable that explained 68.0% of covariance between FC and MBI-C subdomain scores. The latent variable was characterized by high scores across all MBI-C subdomains, indicating global, rather than domain-specific effects of MBI on brain functional networks. Further, the latent variable was associated with whole-brain FC dysfunction between and within networks, particularly in the higher-order default, control and salience/ventral attention networks. ( B-C ) FC matrices display the T-scores of functional connections showing significant (uncorrected P < 0.05) associations (hot colour: positive association; cool colour: negative association) with ( B ) MBI-C total score and ( C ) MBI diagnosis. Both higher MBI-C total score and MBI positivity were associated with widespread within- and between- network FC disruptions notably in higher-order networks, recapitulating the FC dysfunction pattern observed in the partial least squares correlation analysis. MBI = Mild Behavioural Impairment; FC = functional connectivity; MBI-C = Mild Behavioural Impairment Checklist; SalVentAttn = salience/ventral attention; DorsalAttn = dorsal attention; SomMot = somatomotor; TempPar = temporoparietal

Article Snippet: Behaviour partial least squares correlation was then performed on the standardized FC and MBI-C subdomain score residuals using the PLS toolbox [ ] in MATLAB.

Techniques: Functional Assay, Control, Biomarker Discovery