A Candidate Plasma Protein Classifier to Identify Alzheimer's Disease

Article type: Research Article

Authors: Zhao, Xuemei^{a; *} | Lejnine, Serguei^b | Spond, Jeffrey^a | Zhang, Chunsheng^b | Ramaraj, T.C.^c | Holder, Daniel J.^d | Dai, Hongyue^{b; 1} | Weiner, Russell^a | Laterza, Omar F.^{a; *}

Affiliations: [a] Molecular Biomarkers and Diagnostics Laboratory, Merck Research Laboratories, Rahway, NJ, USA | [b] Informatics and Analysis, Merck Research Laboratories, Boston, MA, USA | [c] Biometrics Research, Merck Research Laboratories, Rahway, NJ, USA | [d] Biometrics Research, Merck Research Laboratories, West Point, PA, USA

Correspondence: [*] Correspondence to: Xuemei Zhao, Merck Research Laboratories, RY50A-300, 126 East Lincoln Avenue, Rahway, NJ 07065, USA. Tel.: +1 732 594 1747; Fax: +1 732 594 7804; E-mail: xuemei_zhao@merck.com (Xuemei Zhao); Omar F. Laterza, Merck Research Laboratories, RY50-100, 126 East Lincoln Avenue, Rahway, NY 07065, USA. Tel.: +1 732-594-1486; E-mail: omar_laterza@merck.com (Omar F. Laterza).

Note: [1] Present address: Sage Bionetworks, Seattle, WA, USA and M2Gen, Tampa, FL, USA.

Abstract: Biomarkers currently used in the aid for the diagnosis of Alzheimer's disease (AD) are cerebrospinal fluid (CSF) protein markers and brain neuroimaging markers. These biomarkers, however, either involve semi-invasive procedures or are costly to measure. Thus, AD biomarkers from more easily accessible body fluids, such as plasma, are very enticing. Using an aptamer-based proteomic technology, we profiled 1,129 plasma proteins of AD patients and non-demented control individuals. A 5-protein classifier for AD identification was constructed in the discovery study with excellent 10-fold cross-validation performance (90.1% sensitivity, 84.2% specificity, 87.9% accuracy, and AUC as 0.94). In an independent validation study, the classifier was applied and correctly predicted AD with 100.0% sensitivity, 80.0% specificity, and 90.0% accuracy, matching or outperforming the CSF Aβ42 and tau biomarkers whose performance were assessed in individual-matched CSF samples obtained at the same visit as plasma sample collection. Moreover, the classifier also correctly predicted mild cognitive impairment, an early pre-dementia state of the disease, with 96.7% sensitivity, 80.0% specificity, and 92.5% accuracy. These studies demonstrate that plasma proteins could be used effectively and accurately to contribute to the clinical diagnosis of AD. Although additional and more diverse cohorts are needed for further validation of the robustness, including the support of postmortem diagnosis, the 5-protein classifier appears to be a promising blood test to contribute diagnosis of AD.

Keywords: Alzheimer's disease, biomarkers, blood, aptamers, proteome

DOI: 10.3233/JAD-141149

Journal: Journal of Alzheimer's Disease, vol. 43, no. 2, pp. 549-563, 2015