Research Criteria for the Behavioral Variant of Alzheimer Disease

This systematic review and meta-analysis reviews literature on the behavioral variant of Alzheimer disease and uses the outcomes to propose research criteria for this syndrome.


FDG-PET
The anterior default mode network showed highest goodness-of-fit in bvAD (tAD <bvAD=bvFTD), and significantly less metabolic connectivity of the posterior cingulate cortex to the (right) prefrontal cortex was observed in bvAD compared to tAD eFigure 1. Funnel plots of meta-analyses of behavorial/neuropsychiatric data for behavioral variant AD versus typical AD and bvFTD.

White matter hyperintensities
Funnel plots displaying the position of individual studies on their standardized mean difference (x-axis) relative to their standard error (y-axis).If no publication bias were present, studies would be aligned symmetrically within the dotted triangles, indicating symmetrical locations surrounding the mean effect size, with smaller studies at the lower ends of the plot and larger studies on the higher end of the plot..The dark blue, medium dark blue and light blue parts represent the locations where the effect of the individual study is significant at p<0.05, p<0.025 and p<0.01 compared to the standardized mean difference at 0, whereas the dotted lines represent the mean effect size of the specific studies included.The current plots suggest a lower symmetrical tendency in bvAD vs tAD contrasts compared to bvAD vs bvFTD contrasts, indicating higher publication bias in the bvAD vs tAD contrasts, although the number of studies and sample sizes were small.eFigure 2. Funnel plots of meta-analyses for behavioral and neuropsychiatric symptom data separately for bvAD vs typical AD and bvFTD.
Funnel plots displaying the position of individual studies on their standardized mean difference (x-axis) relative to their standard error (y-axis).If no publication bias were present, studies would be aligned symmetrically within the dotted triangles, indicating symmetrical locations surrounding the mean effect size, with smaller studies at the lower ends of the plot and larger studies on the higher end of the plot.The dark blue, medium dark blue and light blue parts represent the locations where the effect of the individual study is significant at p<0.05, p<0.025 and p<0.01 compared to the standardized mean difference at 0, whereas the dotted lines represent the mean effect size of the specific studies included.The current plots suggest a higher symmetrical tendency in the MMSE contrasts than in the memory and executive domains, indicating higher publication bias in the memory and executive functioning domains than in the MMSE, although the number of studies and sample sizes were small.eFigure 3. Funnel plots of meta-analyses of neuropathological data in bvAD versus typical AD and bvFTD.
Funnel plots displaying the position of individual studies on their standardized mean difference (x-axis) relative to their standard error (y-axis).If no publication bias were present, studies would be aligned symmetrically within the dotted triangles, indicating symmetrical locations surrounding the mean effect size, with smaller studies at the lower ends of the plot and larger studies on the higher end of the plot.The dark blue, medium dark blue and light blue parts represent the locations where the effect of the individual study is significant at p<0.05, p<0.025 and p<0.01 compared to the standardized mean difference at 0, whereas the dotted lines represent the mean effect size of the specific studies included.Although few studies were included per plot, the current plots show an overall symmetrical tendency, marking marginal publication bias.eFigure 4. Summary results of Risk of Bias assessment according to the ROBINS-I tool for studies included in the meta-analyses The ROBINS-I tool for non-randomized studies (https://www.riskofbias.info/)was applied to assess Risk of Bias across studies.Since the domains 'Bias in classification of interventions' and 'Bias due to deviations from intended interventions' were not applicable to the currently assessed studies, these were not filled out (NA=not available).See Table S3 for further details.The figure shows results of meta-analyses across frontal (top row), medial temporal (middle row) and occipital (bottom row) regional quantification of postmortem tau (left column) and amyloid-β (right column) pathology in bvAD versus typical AD.Frontal regions included the frontal pole 4 , middle frontal gyrus 3 and randomly selected frontal areas 2 , and was not further specified in one study 1 .For all meta-analyses, positive standardized mean differences indicate a greater neuropathological burden in bvAD versus typical AD.
SMD=standardized mean difference, S-Q=semi-quantitative.eFigure 8. Differences and overlap between bvAD and dysexecutive AD Differences and overlap between the behavioral variant of AD (bvAD) as proposed in the current work and the dysexecutive variant of AD as proposed elsewhere 85 in terms of behavioral features, cognitive performance and confirmation of AD pathology.

eTable 1 . 2 . 3 . 4 . 5 . 6 . 1 . 2 . 3 . 4 . 5 . 6 .
Full database queries eTable Selection of frontal regions in autopsy studies eTable Risk of bias assessment eTable Characteristics of included studies eTable Percentage of bvFTD features and NPI items in bvAD, bvFTD and tAD eTable Functional connectivity and white matter hyperintensities in bvAD eFigure Funnel plots for behavioral/neuropsychiatric data in meta-analysis eFigure Funnel plots for cognitive data in meta-analysis eFigure Funnel plots for neuropathological data in meta-analysis eFigure Risk of bias assessment summary eFigure Flow chart of study inclusion eFigure Results of meta-analysis for behavioral and neuropsychiatric separately eFigure 7. Meta-analyses for neuropathological data in bvAD vs typical AD

eFigure 6 .eFigure 7 .
Meta-analyses for behavior and neuropsychiatric symptoms separately in bvAD vs typical AD and bvFTDPlots showing meta-analysis results for behavior and neuropsychiatric symptoms separately between patient groups.These plots show similar scores in both behavioral as neuropsychiatric scales scores in bvAD versus bvFTD and a similar difference in behavioral and neuropsychiatric scale scores in bvAD versus typical AD.For all meta-analyses, positive standardized mean differences indicate a greater neuropathological burden in bvAD versus typical AD. bvAD=behavioral variant of AD, tAD=typical AD, bvFTD=behavioral variant frontotemporal dementia, FAB=Frontal Assessment Battery, PBAC=Philadelphia Brief Assessment of Cognition, DCQ=Dépistage Cognitif de Québec, SMD=standardized mean difference.Meta-analyses for neuropathological data in bvAD vs typical AD