Premium
P4–195: Missing data and the identification of optimal predictors of conversion to Alzheimer's disease
Author(s) -
Dowling Maritza
Publication year - 2013
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1016/j.jalz.2013.05.1586
Subject(s) - missing data , discriminative model , identification (biology) , disease , computer science , outcome (game theory) , statistics , data mining , econometrics , artificial intelligence , machine learning , medicine , mathematics , botany , mathematical economics , biology
Background: Older adults with memory-centered and informant-validated cognitive complaints (CC) in the absence of neuropsychological deficits or depression represent a potential at-risk group for progression to amnestic MCI and Alzheimer’s disease (AD). The goals of the present study were to evaluate 1) the rate of conversion of CC participants to MCI over two years and 2) potential antecedent markers associated with future progression to MCI, including baseline cognition, self and informant ratings, genetics, and atrophy on MRI. Amyloid and TSPO/microglial PET and fluid biomarkers were available for a subgroup. Methods: 42 CC participants with baseline and 2-year follow-up data were analyzed from an ongoing 2-site study of memory and aging (Dartmouth and Indiana University). The CognitiveComplaint Index [1]was used to quantitate self and informant ratings. CC participants were categorized as converters to MCI (CC-C) or stable (CC-S) based on clinical consensus. 44 cognitively normal controls (HC) were included for comparison. Baseline structural MRI scans were processed using voxel-based morphometry (VBM) and Freesurfer to extract grey matter density (GMD)andvolumetricmeasures from targeted regions of interest (ROIs). Demographics and medical history, APOE ε4 status, baseline cognitive performance, selfand informant-ratings of cognition, and neuroimaging ROI measures were compared between groupswith age, gender, education and intracranial volume (ICV) included as covariates where appropriate. Results: 10 of 42CC participants (23.8%) converted to early (5) or late (5)MCIwithin 2 years (annualized rate, 12%).At baseline, CC-Cdidnot show consistent differences in cognition from CC-S, except for lower Mattis Dementia Rating Scale total score. Higher self and especially informant-based cognitive complaints at baseline were associated with conversion. Converters compared to CC-S trended toward decreasedMRI volumes andGMDbut this did not reach significance due to insufficient power. Additional imaging and genetic pathway data are being evaluated. Conclusions: Euthymic older adults with elevated levels of cognitive complaints, particularly based on informant ratings, appear to be at-risk for conversion toMCI even in the context of generally normal neuropsychological performance. Additional studies in larger samples are warranted as this this may represent the earliest symptomatic stage of preclinical AD. [1] Saykin Neurology; 2006;67:834-842.