Sexually dimorphic differences in angiogenesis markers are associated with brain aging trajectories in humans

Abstract

Aberrant angiogenesis could contribute to the development of cognitive impairment and represent a therapeutic target for preventing dementia. However, most studies addressing angiogenesis and cognitive impairment focus on model organisms. To test the relevance of angiogenesis to human cognitive aging, we evaluated associations of circulating blood markers of angiogenesis with brain aging trajectories in a pooled two-center sample from deeply phenotyped longitudinal human cohorts (n = 435; female = 207, age = 74 ± 9) using cognitive assessments, biospecimens, structural brain imaging, and clinical data. Blood markers included ligands involved in angiogenesis and vascular function such as basic fibroblast growth factor (bFGF), members of the vascular endothelial growth factor family (VEGFA, VEGFB, and VEGFC), and placental growth factor (PlGF), in addition to their receptors VEGF receptor 1 (VEGFR1) and tyrosine kinase with immunoglobulin and EGF homology domain 2 (Tie2). Machine learning and traditional statistics revealed sexually dimorphic associations of plasma angiogenic growth factors with brain aging outcomes, including executive function and gray matter atrophy. Specifically, markers of angiogenesis were associated with higher executive function and less brain atrophy in younger women (not men), a directionality of association that reversed around age 75. Higher concentrations of bFGF, known for pleiotropic effects on multiple cell types, predicted favorable cognitive trajectories in both women and men. An independent sample from a multicenter dataset (MarkVCID; n = 80; female = 30, age = 73 ± 9) was used to externally validate these findings. In conclusion, this analysis demonstrates the association of angiogenesis to human brain aging, with potential therapeutic implications for vascular cognitive impairment and dementia.

Figure 1.. Unsupervised angiogenesis pattern detection workflow.

Missing values analysis was performed to determine data missing patterns across all observations (n) and variables (p) (a). Missing values were then imputed using multiple imputations by chained equations (MICE) producing m number of imputed datasets, in this study m = 20 (b). The resulting imputed datasets were aggregated using the median for each data point, and a non-linear principal component analysis (NL-PCA) was performed to extract patterns of association or principal components (PC) between variables in a low dimensional space (c). The resulting PC scores were used for further analysis of clinical association by linear mixed models (d). The number of relevant PCs was selected using a permutation-based method. Only PCs from the original NL-PCA (a to c) with information exceeding that of randomly generated PCs, loaded on by more than one variable and stable to missingness were retained (e). PC1, Aberrant Angiogenesis (AA), percent variance explained and loadings of individual factors (f), and PC2, Vascular Health (VH), percent variance explained and loadings of individual factors e (g).

Figure 2.. Demographic associations with angiogenic patterns.

The contribution of markers of interest (PlGF, VEGFR1, bFGF) with each of the selected PCs (PC1, Aberrant Angiogenesis; PC2, Vascular Health) is shown in (a). Arrows pointing to the right (red) represent positive loadings and arrows pointing to the left (blue) represent negative loadings. The relationship of each PC and marker concentrations to age and sex have been studied using linear mixed models (LMM). The graphs show the predicted values (bold lines) and standard errors (shadow ribbons) of PC1 (AA) (b), PC2 (VH) (c), PlGF (d), VEGFR1 (e) and bFGF (f). For each model, the p value for the Age x Sex interaction term and for the Age term is provided.

Figure 3.. Association of angiogenic principal components with a neuroimaging marker of white matter injury.

Linear mixed models (LMMs) were used to study the association of PC1 (AA) with white matter hyperintensity (WMH) when considering the effect of age and sex. To visualize all the effects described by the model, we represent the predicted values of WMH (a) for women and men separately. An illustrative representation of the continuous dependent effect of age on PC1 association with WMH, the predictions for three ages (65, 75 and 85 years) are shown in the line graphs (a). Considering all potential predictions of the model given age creates a prediction surface in a 3D space (b and c). The three ages are shown over the surface for reference. The effect of PlGF on explaining WMH is show in d. The results of the model predicting WMH by bFGF and Age is shown in e. The graphs show the predicted value (bold line or surface) and standard error (shadow ribbon).

Figure 4.. Association of Aberrant Angiogenesis to grey matter volume.

The continuous interaction of AA and age for E4 non-carriers and carriers Women and for Men is shown in a and b respectively. The prediction for three ages (65, 75 and 85 years) are shown in the line graphs. Considering all potential predictions of the model given age creates a prediction surface in a 3D space (c for E4 carrier women; d for E4 carrier men). The subsequent graphs show the predicted value (bold line) and standard error (shadow ribbon) for the continuous interaction of GMV and the individual markers PlGF (e), and VEGFR1 (f for E4 carrier women; g for E4 carrier men). See also Table 1.

Figure 5.. Association of Vascular Health with grey matter volume.

The LMM model testing association of Vascular Health (PC2) with GMV (a). The LMM model testing interaction of bFGF and age with GMV between APOE genotypes in women (b) and men (c). Shadow ribbon shows the standard error.

Figure 6.. Association of executive function with Aberrant Angiogenesis and vascular health.

The LMM model for prediction of executive function for three ages (65, 75 and 85 years) is shown in a, and the surface plot for the continuous interaction of age and AA is shown in fig. S4. The LMM model for the prediction of the executive function by PlGF (b). LMM model for prediction of executive function by VEGFR1, age and sex as predictor (c). The mediation analysis for the mediation of AA and VEGFR1 effect on executive function through GMV is shown in (d). The graphs show the predicted value (bold line) and standard error (shadow ribbon). The prediction of executive function for three ages (65, 75 and 85 years) is shown.

Figure 7.. External validation of Aberrant Angiogenesis loadings and sex-dependent association with executive function.

In an independently collected dataset of 80 aged individuals principle components were derived (a). LLM model showing AA as predictor and executive function as outcome in women and men (b).