Premium
P4‐237: A glyco tmt‐srm assay for measuring plasma clusterin glycosylation associated with increasing brain atrophy in Alzheimer's disease
Author(s) -
Liaing Hui-Chung,
Russell Claire,
Pike Ian,
Ward Malcolm
Publication year - 2015
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.2015.08.066
Subject(s) - clusterin , glycosylation , chemistry , glycoprotein , protein subunit , microbiology and biotechnology , biochemistry , biology , gene , apoptosis
and 125 of the samples passed analytical review (59 AD and 66 controls). Samples were analyzed by a reference laboratory (Becton Dickinson). CD69 (a surface marker of cell cycle activity) was measured on peripheral T, B, and monocyte cells by flow cytometry with and without mitogenic stimulation. Each subject was characterized by multiple measured features that were collected using the LymPro Test methodology. Public and proprietary feature selection algorithms were used to identify optimal feature sets to maximize diagnosis prediction performance using several commonly used machine-learning methods. Prediction performance was compiled using leave-one-out cross-validation measures. Results: Multiple markers were significantly (p<0.05) different in AD subjects compared with HN subjects using univariate models with multiple markers achieving AUCs of 0.657-0.689. Multivariate results indicate that the cross-validated accuracy can be as high as (or even higher than) 80%, especially for methods based on Support Vector Machines approach, even if only 15-25 (optimally selected) LymPro markers are used for the classification. Conclusions: In this analysis, the LymPro Test reports multiple multivariate combinations which discriminate AD from controls with an accuracy of 80% or higher. The blood-based biomarker panel appears to be a robust and reproducible measure of a key pathology of AD and the results corroborate previously published LymPro findings. Accuracy is likely to become more apparent when measured against biomarker defined study subjects. Additional research is planned.