Peripheral innate immunophenotype in neurodegenerative disease: blood-based profiles and links to survival

Abstract

The innate immune system plays an integral role in the progression of many neurodegenerative diseases. In addition to central innate immune cells (e.g., microglia), peripheral innate immune cells (e.g., blood monocytes, natural killer cells, and dendritic cells) may also differ in these conditions. However, the characterization of peripheral innate immune cell types across different neurodegenerative diseases remains incomplete. This study aimed to characterize peripheral innate immune profiles using flow cytometry for immunophenotyping of peripheral blood mononuclear cells in n = 148 people with Alzheimer's disease (AD), frontotemporal dementia (FTD), corticobasal syndrome (CBS), progressive supranuclear palsy (PSP), Lewy body dementia (LBD) as compared to n = 37 healthy controls. To compare groups, we used multivariate dissimilarity analysis and principal component analysis across 19 innate immune cell types. We identified pro-inflammatory profiles that significantly differ between patients with all-cause dementia and healthy controls, with some significant differences between patient groups. Regression analysis confirmed that time to death following the blood test correlated with the individuals' immune profile weighting, positively to TREM2+ and non-classical monocytes and negatively to classical monocytes. Taken together, these results describe transdiagnostic peripheral immune profiles and highlight the link between prognosis and the monocyte cellular subdivision and function (as measured by surface protein expression). The results suggest that blood-derived innate immune profiles can inform sub-populations of cells relevant for specific neurodegenerative diseases that are significantly linked to accelerated disease progression and worse survival outcomes across diagnoses. Blood-based innate immune profiles may contribute to enhanced precision medicine approaches in dementia, helping to identify and monitor therapeutic targets and stratify patients for candidate immunotherapies.

Fig. 1. Euclidean Distance dissimilarity analysis and hierarchical clustering.

Colours and values represent the Euclidean distance calculated from a 19-cell vector averaged across each group. The value for the Euclidean distance is relative to the dataset, thus the number displayed represents the length of the line segment between groups relative to the total group distance. Darker colours and larger Euclidean distance values indicate greater dissimilarity between 19-cell immune profiles across diagnostic groups, while lighter colours indicate relative similarity. The dendrogram represents single linkage hierarchical clustering. Note that all patients separate from controls initially and that the group of syndromes associated with frontotemporal lobar degeneration are similar to each other (PSP, FTD, and CBS), in contrast to Alzheimer’s disease (AD) and Lewy Body dementia (LBD). These relative similarities are highlighted in yellow.

Fig. 2. Immune profile loading onto TREM2+ monocytes and nonclassical monocytes (positively) and classical Monocytes (negatively).

A Median individual loading onto PC1 between controls and all-cause dementia patients (W = 3351, p = 0.02). B Median individual loading onto PC1 for each diagnostic group. Note the absence of significant differences between patient groups. C Correlations of each cell type with PC1. Darker colours indicate stronger positive and negative correlations to PC1. Mann–Whitney U and Kruskal–Wallace tests were used to compare medians. Results were considered significant at p <0.05 indicated with *.

Fig. 3. Immune profile loading onto DC-/- cells (positively) and CCR5+ monocytes (negatively).

A Median individual loading onto PC2 between controls and all-cause dementia patients. B Median individual loading onto PC2 for each group. C Correlations of each cell type with PC2. Darker colours indicate stronger positive and negative correlations to PC3. Mann–Whitney U and Kruskal–Wallace tests were used to compare medians. Results were considered significant at p < 0.05.

Fig. 4. Immune profile loading onto CD16- NK cells (positively) and CD16+ NK cells (negatively).

A Median individual loading onto PC3 between controls and all-cause dementia patients, W = 2025, p = 0.015. B Median individual loading onto PC3 for each group. Post hoc comparisons indicate LBD differs individually from controls, (H = 3.38, p = 0.015). C Cellular correlations to PC3 extracted following PCA. Darker colours indicate stronger positive and negative correlations to PC3. Mann–Whitney U and Kruskal–Wallace tests were used to compare medians followed by Dunn’s Post hoc analysis with FDR correction for multiple comparison. A dashed line indicates a result of statistical significance prior to FDR correction that did not maintain significance following correction.

Fig. 5. Years of survival following blood draw correlate with individual PC1 scores across diagnostic groups.

A linear regression was used to establish how years of survival following blood draw was predicted by individual PC1 loading across all patient groups (β = 0.450, F = 10.098, p = 0.0026, r² = 0.184), in a subset of 42 patients who had died by the census date. Outcomes were considered significant at p < 0.05.