Improving CSF Biomarkers' Performance for Predicting Progression from Mild Cognitive Impairment to Alzheimer's Disease by Considering Different Confounding Factors: A Meta-Analysis

Abstract

Background: Cerebrospinal fluid (CSF) biomarkers' performance for predicting conversion from mild cognitive impairment (MCI) to Alzheimer's disease (AD) is still suboptimal.

Objective: By considering several confounding factors we aimed to identify in which situations these CSF biomarkers can be useful.

Data sources: A systematic review was conducted on MEDLINE, PreMedline, EMBASE, PsycInfo, CINAHL, Cochrane, and CRD (1990-2013).

Eligibility criteria: (1) Prospective studies of CSF biomarkers' performance for predicting conversion from MCI to AD/dementia; (2) inclusion of Aβ42 and T-tau and/or p-tau. Several meta-analyses were performed.

Results: Aβ42/p-tau ratio had high capacity to predict conversion to AD in MCI patients younger than 70 years. The p-tau had high capacity to identify MCI cases converting to AD in ≤24 months.

Conclusions: Explaining how different confounding factors influence CSF biomarkers' predictive performance is mandatory to elaborate a definitive map of situations, where these CSF biomarkers are useful both in clinics and research.

Keywords: Alzheimer’s disease; CSF biomarkers; confounding factors; meta-analysis; mild cognitive impairment; systematic review.

Figure 1

Study selection flow.

Figure 2

Positive and negative likelihood ratios. A LR+ greater than one increases the pretest probability that the disease is present [in this context progression from MCI to AD or, in other words, MCI due to AD (Albert et al., 2011)]. A LR– of less than one diminishes the pretest probability that disease is present. The established guidelines (Qizilbash, 2002) states that a LR+ greater than 10 will often make conclusive changes to the pretest probability, indicating that the disease is likely present; a LR+ between 5 and 10 corresponds to moderate increase in probability; and a LR+ between 2 and 5 corresponds to small increase. A LR− of less than 0⋅1 will often make conclusive changes to the pretest probability that the disease is present, indicating that the disease is unlikely present; a LR− between 0⋅1 and 0⋅2 corresponds to moderate decrease in probability; and a LR− between 0⋅2 and 0⋅5 corresponds to small decrease. LR+, positive likelihood ratio; LR−, negative likelihood ratio; Global MA, global meta-analysis; MCI, mild cognitive impairment; AD, Alzheimer’s disease.