Background
The vascular endothelial growth factor (VEGF) family is involved in angiogenic signaling and has been implicated in Alzheimer’s Disease (AD). Previous work from our team has found that ...higher transcript levels of VEGFB, PGF, FLT1, and FLT4 in the brain relate to worse cognitive performance and higher levels of AD neuropathology. This study explored whether comparable associations are observed when measuring mRNA in whole blood.
Method
Transcriptomic, cognitive, and biomarker data were acquired from the Vanderbilt Memory & Aging Project. Participants (n=335) included older adults (age=72+/‐7.32) with normal cognition (n=203) or mild cognitive impairment (n=132). Expression of the ten genes in the VEGF family were quantified using RNA sequencing of blood. Outcomes included a composite measure of memory and cerebrospinal fluid (CSF) levels of β‐amyloid, total tau, and phosphorylated tau (ptau181). Linear regression models related VEGF family member expression to baseline cognitive performance and AD biomarkers. Linear mixed effects regression models assessed longitudinal change in cognition. All models covaried for age at baseline and sex. To aid in interpretation, Pearson correlations between blood and brain expression were calculated leveraging data from 137 participants from the Genotype Tissue Expression Project (GTexportal.org).
Result
At baseline, higher levels of VEGFB mRNA were associated with better memory performance (Figure 1, β=0.47, p=0.008). In longitudinal analyses, higher baseline expression of PGF was associated with more rapid decline in memory (β= 0.02, p=0.04). No other significant associations were observed (p‐values>0.09). Among GTEx participants, higher expression of VEGFB in blood was correlated with lower expression in the brain (Figure 2, r=‐0.19, p=0.02). PGF correlations did not reach statistical significance (r=0.09, p=0.27).
Conclusion
Higher PGF expression in blood relates to worse cognitive outcomes, consistent with previous findings in the brain. In contrast, higher levels of VEGFB in blood relates to better performance, opposite of the previously observed association in the brain. Notably, in GTEx we observed opposing correlations between blood levels of VEGFB and brain levels of VEGFB, suggesting that higher levels of VEGFB in the blood may indeed reflect lower levels in the brain. Our results highlight the potential of VEGFB and PGF as blood biomarkers in aging and AD.
Background
Alzheimer’s disease (AD) polygenic risk scores (PRS) are often derived from case/control GWAS, which are typically not sensitive to preclinical disease changes, limiting their clinical ...utility. To overcome this pitfall, we built and evaluated the performance of multiple PRS of AD‐related endophenotypes, including resilience to cognitive impairment in the presence of amyloid.
Method
Four PRS were derived from GWAS of baseline memory, memory decline, cognitive resilience (Dumitrescu et al., 2021), and AD (Kunkle et al. 2019). PRS were built in an independent cohort that was free of dementia, the Vanderbilt Memory and Aging Project (N=255). We used linkage disequilibrium clumping on TOPMed‐imputed genotypes, and a threshold of P=0.01. We ran linear models with baseline memory score as the outcome, and baseline age, sex, and PRS as predictors. Linear mixed effects models, with identical predictors, were used when longitudinal memory was the outcome, letting individual slope and intercept vary. Finally, we tested a CSF Aβ42‐by‐PRS interaction term to assess if PRS modified associations between amyloid and cognition. Sensitivity analyses excluded APOE from PRS calculations.
Result
As expected, the baseline memory PRS related to baseline memory (β=0.12, P=0.04), and the memory decline PRS related to longitudinal memory (β=0.04, P=0.001). Without APOE, the baseline memory PRS and the memory decline PRS results attenuated. The AD PRS related to both baseline memory (β=‐0.19, P=0.002) and longitudinal memory (β=‐0.03, P=0.01), but results attenuated without APOE. The resilience PRS did not have a main effect on baseline or longitudinal memory. However, interestingly, the resilience PRS interacted with CSF Aβ42 on baseline memory (β=‐0.001, P=0.02; Figure 1), whereby the resilience PRS related to memory among amyloid‐positive individuals (β=0.44, P=0.01) but not amyloid‐negative individuals (β=0.06, P=0.46). No other PRS was predictive of memory among amyloid‐positive individuals, and resilience PRS results remained consistent without APOE.
Conclusion
Our results demonstrate that a resilience PRS appears to be predictive of individual cognitive performance downstream of amyloidosis. Future work is needed to replicate this finding, but our preliminary findings highlight the potential utility of resilience PRS for predicting individual risk for AD‐related cognitive impairment during the preclinical stages of disease.
Background
Cerebrospinal fluid (CSF) soluble triggering receptor expressed on myeloid cells‐2 (sTREM2) is an emerging biomarker of neuroinflammation in Alzheimer’s disease (AD). Yet, sTREM2 ...expression has not been systematically evaluated in relation to concomitant drivers of neuroinflammation. We previously found a novel association between high Aβ40 (but not Aβ42) and high sTREM2 levels at baseline and replicated additional associations between a fluid biomarker of blood‐brain barrier (BBB) integrity and CSF biomarkers of neurodegeneration. Therefore, we sought to determine the relative contributions of these biological correlates to CSF sTREM2 levels during the prodromal stages of AD. Additionally we investigated whether the residual variance in sTREM2 level correlated with each biomarker predicted future cognitive performance.
Method
Leveraging 155 Vanderbilt Memory and Aging Project (VMAP) participants (mean±standard deviation age=72±6, male=67%, mild cognitive impairment=47%), previously established associations between baseline CSF sTREM2 concentrations and CSF AD biomarkers (Aβx‐40, total tau, phosphorylated tau, and neurofilament light) and BBB integrity (CSF/plasma albumin ratio) were examined via hierarchical linear regression using sTREM2 as an outcome variable. Models adjusted for age, sex, education, cognitive status, and APOE‐ε4 status. Baseline CSF sTREM2 levels and corresponding residual components when adjusting for other biomarker levels in independent models were used in longitudinal mixed effects models predicting an episodic memory composite calculated as a z‐score from independent tests. Participants had up to five visits (mean±sd=2.63±1.30 visits).
Result
High baseline CSF sTREM2 levels predicted slower longitudinal memory decline in VMAP (β=1.442e‐05, p=0.035). Additionally, the component of sTREM2 signal correlated with Aβx‐40 best predicted future memory performance (β=2.217e‐05, p=0.007). Together, Aβx‐40, phosphorylated tau, and the CSF/plasma albumin ratio jointly explained 36% of variance in sTREM2 levels. Aβx‐40 levels and the CSF/plasma albumin ratio explained sTREM2 signal above and beyond established associations with tau (R2=0.21; R2=0.21; and R2=0.17, respectively).
Conclusion
We highlight potential contributions of Aβ homeostasis and BBB integrity to elevations in CSF sTREM2, underscoring novel biomarker associations relevant to disease progression. Results suggest that the tight correlation between baseline CSF sTREM2 and Aβx‐40 levels may have notable importance to cognitive trajectory in an aged cohort enriched for mild cognitive impairment.
•We describe a novel association between blood VEGF expression levels and memory.•We describe a novel association between blood VEGF protein levels and AD pathology biomarkers.•Expression and ...proteins levels of VEGF in blood are potential biomarkers of AD.
The vascular endothelial growth factor (VEGF) family of genes has been implicated in the clinical development of Alzheimer's Disease (AD). A previous study identified associations between gene expression of VEGF family members in the prefrontal cortex and cognitive performance and AD pathology. This study explored if those associations were also observed in the blood. Consistent with previous observations in brain tissue, higher blood gene expression of placental growth factor (PGF) was associated with a faster rate of memory decline (p=0.04). Higher protein abundance of FMS-related receptor tyrosine kinase 4 (FLT4) in blood was associated with biomarker levels indicative of lower amyloid and tau pathology, opposite the direction observed in brain. Also, higher gene expression of VEGFB in blood was associated with better baseline memory (p=0.008). Notably, we observed that higher gene expression of VEGFB in blood was associated with lower expression of VEGFB in the brain (r=-0.19, p=0.02). Together, these results suggest that the VEGFB, FLT4, and PGF alterations in the AD brain may be detectable in the blood compartment.
Cerebrospinal fluid (CSF) levels of amyloid-β 42 (Aβ42) and tau have been evaluated as endophenotypes in Alzheimer’s disease (AD) genetic studies. Although there are sex differences in AD risk, sex ...differences have not been evaluated in genetic studies of AD endophenotypes. We performed sex-stratified and sex interaction genetic analyses of CSF biomarkers to identify sex-specific associations. Data came from a previous genome-wide association study (GWAS) of CSF Aβ42 and tau (1527 males, 1509 females). We evaluated sex interactions at previous loci, performed sex-stratified GWAS to identify sex-specific associations, and evaluated sex interactions at sex-specific GWAS loci. We then evaluated sex-specific associations between prefrontal cortex (PFC) gene expression at relevant loci and autopsy measures of plaques and tangles using data from the Religious Orders Study and Rush Memory and Aging Project. In Aβ42, we observed sex interactions at one previous and one novel locus: rs316341 within
SERPINB1
(
p
= 0.04) and rs13115400 near
LINC00290
(
p
= 0.002). These loci showed stronger associations among females (
β
= − 0.03,
p
= 4.25 × 10
−8
;
β
= 0.03,
p
= 3.97 × 10
−8
) than males (
β
= − 0.02,
p
= 0.009;
β
= 0.01,
p
= 0.20). Higher levels of expression of
SERPINB1, SERPINB6,
and
SERPINB9
in PFC was associated with higher levels of amyloidosis among females (corrected
p
values < 0.02) but not males (
p
> 0.38). In total tau, we observed a sex interaction at a previous locus, rs1393060 proximal to
GMNC
(
p
= 0.004), driven by a stronger association among females (
β
= 0.05,
p
= 4.57 × 10
−10
) compared to males (
β
= 0.02,
p
= 0.03). There was also a sex-specific association between rs1393060 and tangle density at autopsy (
p
female
= 0.047;
p
male
= 0.96), and higher levels of expression of two genes within this locus were associated with lower tangle density among females (
OSTN p
= 0.006;
CLDN16 p
= 0.002) but not males (
p
≥ 0.32). Results suggest a female-specific role for
SERPINB1
in amyloidosis and for
OSTN
and
CLDN16
in tau pathology. Sex-specific genetic analyses may improve understanding of AD’s genetic architecture.
Abstract
Background
It is well‐established that telomeres shrink over the course of aging and that this process is genetically regulated. Work from our group and others has highlighted the complex ...interplay between measured leukocyte telomere length and Alzheimer’s Disease (AD) biomarkers over the course of AD. We expand on this work to explore whether a polygenic score predicting telomere length relates to PET measures of brain amyloidosis in the preclinical stages of disease.
Method
A Polygenic Risk Score (PRS) of telomere length was built from a published GWAS on telomere length from UK Biobank (N = 472,174). We leveraged genetic data from non‐Hispanic White participants from the Alzheimer’s Disease Neuroimaging Initiative (ADNI; N = 1,414) and the Anti‐Amyloid Treatment in Asymptomatic Alzheimer’s study (A4; N = 3,002). The relation of the score to measured telomere length was validated using leukocyte telomere length data from ADNI. A linear regression model was used to evaluate whether the PRS relates to a PET measure of brain amyloidosis in a harmonized dataset including ADNI and A4, with a total sample size of 3,940 after restricting to individuals with both genetic data and amyloid measurements. Covariates included age, sex, and diagnostic status.
Result
As expected, telomere PRS related to measured telomeres in ADNI when covarying for age, sex, and baseline diagnosis (Figure 1, p = 0.045). Interestingly, we also observed a counter intuitive association between genetically predicted telomere length and amyloidosis whereby longer predicted telomere length was associated with higher amyloid burden (Figure 2, β = 1305, p = 0.006).
Conclusion
A PRS for telomere length modestly relates to measured telomere length in older adults and relates to higher amyloid burden. Previous work from our group has highlighted the interaction between measured telomere length and amyloid status such that the association is counter intuitive among those who are biomarker positive. The present results highlight the exciting possibility that telomere length may relate directly to AD biomarkers and open the opportunity for future work further probing the complex interplay between telomere length and AD biomarkers leveraging this powerful genetic tool.
Abstract
Background
Alzheimer’s disease (AD) disproportionately affects women, who make up two‐thirds of all clinical cases of AD. While sex differences in AD neuropathology, the response to ...pathology, and the genetic predictors of clinical AD have been well described, sex differences in the brain transcriptomic signatures of AD endophenotypes have not been fully characterized.
Methods
We leveraged bulk RNA‐sequencing data from three brain regions (dorsolateral prefrontal cortex (DLPFC), posterior cingulate cortex, and caudate nucleus from the Religious Orders Study and Rush Memory and Aging Project (ROS/MAP). Propensity scoring was used to match male to female participants due to disproportionate sample size. As a result, there were 791 participants (50% male; mean age at death = 87.6 years). Sex‐stratified and sex‐interaction (int) regression models assessed sex‐specific transcript associations with amyloid and tau burden, along with longitudinal global cognition. Age at death, latency to death, and post‐mortem interval were included as covariates. Sex‐specific genes were defined as related to a trait in one sex (FDR‐corrected P<0.05) but not in the other sex and that showed evidence of a sex‐modifying effect (P
int
<0.05).
Results
Of the more than 20,000 significant autosomal gene expression associations with the three AD endophenotypes, 11% were sex‐specific with DLPFC having largest number of sex‐specific genes (7%). Despite equivalent percentages of males and females, we observed more female‐specific effects (10%) than male‐specific effects (1%) (
Figure 1
). Several genes showed particularly strong effects among females, including LRIG3 with amyloid (P
men
= 0.20, P
women
= 3.34×10
−06
, P
int
= 0.017) and tau tangles (P
men
= 0.34, P
women
= 0.008, P
int
= 0.008), and SLC22A17 with amyloid (P
men
= 0.8, P
women
= 1.63×10
−04
, P
int
= 0.02) and tau tangles (P
men
= 0.18, P
women
= 4.81×10
−05
, P
int
= 0.04). There were 16 genes with male‐specific associations with at least two traits including FAIM2 (P
men
= 9.9×10
−4
, P
women
= 0.49, P
int
= 0.004), ATP13A2 (P
men
= 8.3×10
−5
, P
women
= 0.68, P
int
= 0.002), and PRSS35 (P
men
= 7.5×10
−5
, P
women
= 0.64, P
int
= 0.02) (
Figure 2
).
Conclusion
Our results highlight sex‐specific transcriptomic associations with AD endophenotypes, including genes for protein misfolding and neurotrophic signaling among females and for neuron apoptotic involvement among males. Similar sex‐specific patterns with AD were observed in independent AMP‐AD cohorts for LRIG3, SLC22A17, and FAIM2 (
https://agora.adknowledgeportal.org/
).These findings highlight the exciting potential of precision medicine approaches that consider sex‐specific biological pathways to select new targets for mechanistic evaluation.
Abstract
Background
We harmonized composite scores for memory, executive functioning (EF), and language from granular cognitive data from the National Alzheimer’s Coordinating Center (NACC) uniform ...dataset on 39,965 individuals. We explored if our cognitive scores correlated with APOE genotype and structural magnetic resonance imaging data.
Method
In our previously published harmonization methods we used confirmatory factor analysis models, guided by theoretical considerations from content experts, to estimate domain scores. For this cross‐sectional analysis of last visits in participants over age 60, we obtained APOE genotype data (n = 28,558 individuals) and FreeSurfer regional volume data (n = 2,404 individuals). We looked at relationship between cognitive performance and APOE genotype grouping by disease group (normal cognition, mild cognitive impairment (MCI), or AD dementia). We ran student’s two sample t‐test to assess the effect of APOE ε4 genotype on cognitive domain score. We also ran regression models for cognitive domain scores on each FreeSurfer region selected a priori on the basis of assumed association on that index domain adjusting for age, gender and APOE genotype.
Result
Relevant demographic and clinical data are summarized in Table 1. Associations with APOE genotype and each cognitive domain are shown in Figure 1. As expected, participants with at least one APOE ε4 alleles compared with none had lower composite memory scores (p<0.01, t‐test) compared with those with none in participants with dementia and MCI. APOE ε4 carriers also had lower language scores in participants with dementia (p = 0.037, t‐test). Among the FreeSurfer regions, composite memory scores were strongly associated with hippocampal volume, parahippocampus and entorhinal cortex thickness across all dementia categories (Table 2). EF scores were associated with 2 EF brain regions of interest in participants with dementia. Language scores were associated with 2 language regions of interest in participants with dementia.
Conclusion
Our analyses shows composite scores are associated with APOE genotype in participants with dementia and relevant brain regions in the memory domain across all dementia categories. Our analyses shows that our cognitive scores correlate with memory brain regions and APOE genotype in expected directions.
Background
Recent decades have brought a renaissance in understanding lysosomes and autophagy in neurobiology and neurodegenerative diseases. Neuronal endolysosomes are defective in Alzheimer’s ...disease (AD), most notably in dystrophic neurites where large numbers of enzymatically inactive lysosomes accumulate, distending axons around β‐amyloid plaques. Granulovacuolar degeneration bodies (GVDBs) are another well‐documented neuropathology associated with AD. These pathological neuronal organelles have two membranes, the outer containing lysosome‐associated membrane proteins suggesting these organelles are a product of frustrated autophagy. To implicate GVDBs in the pathogenesis of AD, we evaluated if GVDBs are present in dystrophic neurites and whether the expression of key GVDB genes are correlated with cognition in AD.
Methods
Human brain with AD and brains from 5xFAD mice were examined with immunohistochemistry and confocal microscopy for the established GVDB marker CK1δ. We also evaluated the recently discovered GVDB marker Golgin a4 and its binding partner ARL1 (implicated in the formation of autophagosomes). Finally, we assessed the association of mRNA levels of these genes in frontal cortex at autopsy with cognition and β‐amyloid pathology in 600 participants in Rush University’s Religious Orders Study and Memory and Aging Project (ROS/MAP).
Results
We discovered that CK1δ, Golgin 4a and ARL1 accumulate in dystrophic neurites in mouse and human AD brain, partially co‐localizing with the lysosomal enzyme cathepsin B, indicating a portion of the organelles accumulating in dystrophic neurites are likely GVDBs. In ROS/MAP participants, golgin a4 expression inversely correlated with cognitive trajectory (β=‐0.003, p=6.0x10‐4) and β‐amyloid plaque density (β=0.01, p=0.001), while ARL1 directly correlated with longitudinal cognition (β=0.004, p=7.0x10‐4) and β‐amyloid plaque density (β=‐0.01, p=0.027) in a model covarying for age at death, sex, education, APOE genotype, and interval between last visit and death. From these results we constructed a conceptual model of the molecular interaction between dysfunctional lysosomes and dysfunctional autophagic organelles in dystrophic neurites.
Conclusions
GVDBs accumulate in dystrophic neurites in AD, indicating the presence of defective autophagic organelles in addition to the known dysfunctional lysosomes. Transcriptomic data demonstrating a correlation of GVDB markers with cognition and β‐amyloid pathology suggests this discovery is likely functionally relevant.