Leukocyte surface biomarkers implicate deficits of innate immunity in sporadic Alzheimer's disease

Abstract

Introduction: Blood-based diagnostics and prognostics in sporadic Alzheimer's disease (AD) are important for identifying at-risk individuals for therapeutic interventions.

Methods: In three stages, a total of 34 leukocyte antigens were examined by flow cytometry immunophenotyping. Data were analyzed by logistic regression and receiver operating characteristic (ROC) analyses.

Results: We identified leukocyte markers differentially expressed in the patients with AD. Pathway analysis revealed a complex network involving upregulation of complement inhibition and downregulation of cargo receptor activity and Aβ clearance. A proposed panel including four leukocyte markers - CD11c, CD59, CD91, and CD163 - predicts patients' PET Aβ status with an area under the curve (AUC) of 0.93 (0.88 to 0.97). CD163 was the top performer in preclinical models. These findings have been validated in two independent cohorts.

Conclusion: Our finding of changes on peripheral leukocyte surface antigens in AD implicates the deficit in innate immunity. Leukocyte-based biomarkers prove to be both sensitive and practical for AD screening and diagnosis.

Keywords: Alzheimer's disease; CD11c; CD163; CD59; CD91; ROC curve; complement system; innate phagocytosis; leukocyte biomarker; peripheral immune response; receptor for advanced glycation end products (RAGE).

FIGURE 1

Differential expression of the leukocyte markers in AD. (A) A Heatmap of the expression levels of the top 27 down‐ and upregulated AD‐associated leukocyte markers in CN, MCI, and AD‐dementia. Blue and red represent a relative color scheme uses the minimum and maximum values in each row to convert values to colors. (B) A heatmap showing the pairwise correlations among AD‐associated leukocyte markers. Each row and column represent one of the 27 AD‐associated leukocyte markers. Red and blue indicate positive and negative correlations between pairs, respectively. Black squares denote the four clusters based on hierarchical clustering, and numbers in brackets on the right indicate the cluster number. (C) A biplot showing both the loadings and the PC scores. Variables in cluster 1 and 2 are strongly correlated (values close to 1 or ‐1) with PC1; while cluster 3 correlates strongly with PC2. (D) Representative GO term enrichment of the 12 AD‐associated leukocyte markers. The GO terms of the down‐ and upregulated leukocyte markers are indicated in blue and red, respectively.

FIGURE 2

Associations between the top two AD‐associated leukocyte markers and AD clinical stages, brain Aβ burden, and cognitive function. AD‐associated leukocyte markers, CD59 and CD91, are presented in A‐D and E‐F, separately. The number of participants is 200 (CN = 124, MCI = 43, and AD = 33). Bar graphs are mean ± standard deviation of the mean and p value is decided by one‐way ANOVA followed by multiple comparison using Tukey honest significant difference (HSD). Correlation r and p values are decided by Pearson product‐moment correlational analysis. If the p‐value is less than or equal to the significance level (alpha = 0.05), the correlation is deemed significantly different to zero, and the magnitude of the correlation is given to talk to the strength of the Pearson correlation R value (low absolute R value < 0.25, medium absolute R value of between 0.25 and 0.75, and high absolute R value of > 0.75. ***: p < 0.001; ****: p < 0.0001

FIGURE 3

Performance assessment of the top four AD‐associated leukocyte markers. (A) Paired sample area difference under the ROC curves between Base model 1 and a composite panel, New 1, when predicting diseased state from non‐diseased state. Base model 1 comprises of age, sex, years of education, and APOE ε4 allele status. New 1 is formed by adding MFI of CD59 on T&B (BM4), MFI of CD163 on CD14^‐ neutrophils (BM8), MFI of CD91 on CD14⁺CD16⁺ monocytes (BM10) and MFI of CD11c on CD14⁻CD16⁺ monocytes (lnBM2) to Base model 1. The New 1 panel proved an AUC of 0.93 (CI: 0.88 to 0.97), improved from the Base model 1 AUC of 0.88 (CI: 0.82 to 0.94) with a p value of 0.021 (DeLong test). (B) Assessing the improvement in risk prediction of New 1 model over Base 1 model. The difference between two models are shown by the filled areas. Light blue and dark blue represent the improvements with (TRUE) and without (FALSE) the event of interest.

FIGURE 4

Scientific hypothesis and molecular pathways inhibiting phagocytosis and complement. Left. Aβ cross‐membrane transportation is augmented given that both CD91 and RAGE increased in AD; Middle. A schematic phagocyte shows down‐regulation in phagocytic receptors, MerTK and P2X7; scavenger receptors, CD36 and CD163; and cell adhesion integrins, CD11b, CD11c, and CD18, which can form CR3 and CR4; Right (adapted from KEGG map04610). Complement pathways are activated by antigen‐antibody complex. It triggers cascade to form C3b. C3b binds to CR1 thatis upregulated in AD, and subsequentlystimulates phagocytosis. Whilst, iC3b binds to CR3 and CR4 thatare downregulated in AD, and subsequently   inhibit phagocytosis. On the other hand, C5b, C6, C7, C8, and C9 form the membrane attack complex (MAC). CD59 prevents formation of MAC and lysis of target cells. Both inhibit opsonization, phagocytosis and clearance of pathogens which are involved in AD pathogenesis and development.