Mengxue Wang,1,* Zhengsheng Zhang,1,* Yachen Shi,2 Hao Shu,1 Chunming Xie,1 Qingguo Ren,1 Zan Wang1 

1Department of Neurology, Zhongda Hospital, School of Medicine, Jiangsu Provincial Key Laboratory of Brain Science and Medicine, Southeast University, Nanjing, 210009, People’s Republic of China; 2Department of Neurology, Nanjing Medical University Affiliated Wuxi People’s Hospital, Wuxi, People’s Republic of China

*These authors contributed equally to this work

Correspondence: Zan Wang, Department of neurology, Zhongda Hospital, School of Medicine, Jiangsu Provincial Key Laboratory of Brain Science and Medicine, Southeast University, Nanjing, 210009, People’s Republic of China, Email ujs_yingxiang0402@126.com

Objective: Patients with amnestic mild cognitive impairment (aMCI) are thought to be highly susceptible to developing Alzheimer’s disease (AD). The study aimed to investigate the possibilities of plasma-biomarkers for individual patient identification of aMCI and prediction of episodic memory.
Methods: We recruited 87 healthy controls and 68 aMCI patients in this study; and 22/68 aMCI patients completed 3-year follow-up visits, with six aMCI patients converting to AD. An ultrasensitive quantitative method was employed to measure the levels of plasma biomarkers.
Results: Relative to healthy controls, the aMCI patients showed significantly higher levels of plasma neurofilament light (NfL) and lower levels of plasma Aβ 40, Aβ 42 and Aβ 42/Aβ 40 ratio (all P values < 0.01). Using multivariate relevance vector regression models, we further demonstrated plasma biomarkers could accurately predict baseline Rey’s Auditory Verbal Learning Test-20 min delayed recall (AVLT-DR) scores (r = 0.362, P value < 0.001) and 3-year longitudinal AVLT-DR changes (r = 0.365, P value < 0.001) for individual aMCI patients; plasma-indicators contributed most to the predictions including total-tau and NfL. Finally, by using support vector machine model, the combination of plasma Aβ 42/Aβ 40, mini-mental state examination (MMSE) score, and hippocampal/parahippocampal volume had the highest accuracy of 77.42% (sensitivity = 72.06%, specificity = 81.61%) for identifying aMCI patients.
Conclusion: We provided support to the use of plasma total-tau and NfL as simple biomarkers to predict the severity of episodic memory deficit for individual aMCI patients and aMCI progression, and further demonstrated that the combination of plasma Aβ 42/Aβ 40, hippocampal/parahippocampal volume, and MMSE score could serve as an integrated screening tool to select aMCI individuals.

Keywords: Alzheimer’s disease, amnestic mild cognitive impairment, episodic memory, machine learning, plasma-biomarkers