Genetic-based patient stratification in Alzheimer's disease

Abstract

Alzheimer's disease (AD) shows a high pathological and symptomatological heterogeneity. To study this heterogeneity, we have developed a patient stratification technique based on one of the most significant risk factors for the development of AD: genetics. We addressed this challenge by including network biology concepts, mapping genetic variants data into a brain-specific protein-protein interaction (PPI) network, and obtaining individualized PPI scores that we then used as input for a clustering technique. We then phenotyped each obtained cluster regarding genetics, sociodemographics, biomarkers, fluorodeoxyglucose-positron emission tomography (FDG-PET) imaging, and neurocognitive assessments. We found three clusters defined mainly by genetic variants found in MAPT, APP, and APOE, considering known variants associated with AD and other neurodegenerative disease genetic architectures. Profiling of these clusters revealed minimal variation in AD symptoms and pathology, suggesting different biological mechanisms may activate the neurodegeneration and pathobiological patterns behind AD and result in similar clinical and pathological presentations, even a shared disease diagnosis. Lastly, our research highlighted MAPT, APP, and APOE as key genes where these genetic distinctions manifest, suggesting them as potential targets for personalized drug development strategies to address each AD subgroup individually.

Figure 1

Overview of the presented methodology and results.

Figure 2

Heatmap showing significantly different edges between clusters. Rows and columns represent gene–gene interaction scores (edge scores) and samples, respectively. Lower values indicate more edge affectation, i.e., more genetic variants in edge interactors. Bottom colors correspond to the obtained clusters: red (Cluster 1), green (Cluster 2), and blue (Cluster 3). Heatmap was generated with seaborn (v.0.12.2).

Figure 3

Genetic variants ranking in the classification of each cluster vs. the rest. Features importances were obtained through Random Forest, using as positive class: (a) Cluster 1, (b) Cluster 2, (c) Cluster 3.

Figure 4

Average metabolism coefficients against controls in each cluster. (a) Baseline, and (b) longitudinal analysis. ROIs correspond to the AAL atlas. The shown regions obtained a p-value (FDR corrected) < 0.05.

Figure 5

ADAS-Cog neurocognitive networks results for each cluster. (a) Neurocognitive networks built for each cluster; nodes are colored according to the cognitive domain that the item or test they represent. Edges are colored and weighted according to the edge weight. (b) Degree centrality of nodes in clusters’ cognitive networks. Edges weights and degree centrality values were obtained through bootstrap calculation (n = 10, 250 repeats).