Plasma proteomic biomarkers and the association between poor cardiovascular health and incident dementia: The UK Biobank study

Abstract

Background: The study examined how plasma proteome indicators may explain the link between poor cardiovascular health (CVH) and dementia risk.

Methods: The present study involved 28,974 UK Biobank participants aged 50-74y at baseline (2006-2010) who were followed-up for ≤ 15 y for incidence of dementia. CVH was calculated using Life's Essential 8 (LE8) total scores. The scores were standardized and reverse coded to reflect poor CVH (LE8z_rev). OLINK proteomics was available on this sample (k = 1,463 plasma proteins). The study primarily tested the mediating effects of the plasma proteome in LE8z_rev-dementia effect. The total effect was decomposed into "mediation only" or pure indirect effect (PIE), "interaction only" or interaction referent (INTREF), "neither mediation nor interaction" or controlled direct effect (CDE), and "both mediation and interaction" or mediated interaction (INTMED).

Results: The study found poorer CVH assessed by LE8z_rev increased the risk of all-cause dementia by 11 % [per 1 SD, hazard ratio, (HR) = 1.11, 95 % CI: 1.03-1.20, p = 0.005). The study identified 11 plasma proteins with strong mediating effects, with GDF15 having the strongest association with dementia risk (per 1 SD, HR = 1.24, 95 % CI: 1.16, 1.33, P < 0.001 when LE8z_rev is set at its mean value) and the largest proportion mediated combining PIE and INTMED (62.6 %; 48 % of TE is PIE), followed by adrenomedullin or ADM. A first principal component with 10 top mediators (TNFRSF1A, GDF15, FSTL3, COL6A3, PLAUR, ADM, GFRAL, ACVRL1, TNFRSF6B, TGFA) mediated 53.6 % of the LE8z_rev-dementia effect. Using all the significant PIE (k = 526) proteins, we used OLINK Insight pathway analysis to identify key pathways, which revealed the involvement of the immune system, signal transduction, metabolism, disease, protein metabolism, hemostasis, membrane trafficking, extracellular matrix organization, developmental biology, and gene expression among others. STRING analysis revealed that five top consistent proteomic mediators were represented in two larger clusters reflecting numerous interconnected biological gene ontology pathways, most notably cytokine-mediated signaling pathway for GDF15 cluster (GO:0019221) and regulation of peptidyl-tyrosine phosphorylation for the ADM cluster (GO:0050730).

Conclusion: Dementia is linked to poor CVH mediated by GDF15 and ADM among several key proteomic markers which collectively explained ∼ 54 % of the total effect.

Keywords: Aging; Cardiovascular health; Dementia; Life’s essential 8; Plasma proteomic biomarkers.

FIGURE 1.. Poor cardiovascular health (CVH; reversed tertiles of LE8) vs. all-cause dementia, overall and by sex: UK biobank 2006–2021

Abbreviations: chi2=chi-square test; CI=confidence interval; KM=Kaplan Meier; LE8=Life’s Essential 8; Pr=P-values; T1=First tertile (ideal CVH); T2=Second teritle (intermediate CVH); T3=Uppermost tertile (Poor CVH); UK=United Kingdom. Note: Chi2 refers to a log-rank test. Details are provided on github: https://github.com/baydounm/UKB-paper11-supplementarydata.

FIGURE 2.. Volcano plot of plasma proteomic biomarkers in relation to LE8 reverse-coded z-scored total score: UK biobank 2006–2010

Abbreviations: See list of abbreviations for protein abbreviations; LE8=Life’s Essential 8. Note: Based on a series of multiple linear regression models, with main predictor being LE8 toal score (z-scored, reverse-coded by multiplying by −1) and the outcome being each of 1,463 plasma proteomic biomarkers (Log2 transformed, z-scored). The y-axis is the predictor’s associated p-value on a −Log10 scale and the X-axis is the β coefficient (effect of LE8 exposure on standardized z-scores of plasma proteomic markers) from the multiple linear regression models. An estimate with a Bonferroni corrected p-value<0.05 and an absolute estimate >0.20 is marked by the plasma proteomic marker abbreviation (See UKB showcase URL: https://biobank.ndph.ox.ac.uk/showcase/ ). Selected proteins (k=144) for further mediation analysis have a corrected p-value<0.05 and a point estimate >0.20 in absolute value (red). Dashed vertical lines are set at −0.20 and +0.20 effect sizes. Dashed horizontal line is set at the Bonferroni corrected p-value. Details are provided on github: https://github.com/baydounm/UKB-paper11-supplementarydata.

FIGURE 3. Four-way decomposition of the association between poor cardiovascular health as measured by LE8 total score (z-scored, reverse coded) and incidence of all-cause dementia by the selected plasma proteomic biomarkers (k=144): UK biobank 2006–2021

(A) Heatmap for raw 4-way decomposition: PIE p>0.05 (no significant pure mediation), k=77 (B) Heatmap for raw 4-way decomposition: PIE p<0.05, PIE<0, TE0, TE>CDE (consistent mediation), k=11 Note: Mediation Effects could range from 0 to 0.5. Abbreviations: CDE= excess relative risk due to neither mediation nor interaction or controlled direct effect; INTMED= excess relative risk due to mediated interaction or mediated interaction; INTREF= excess relative risk due to interaction only or interaction referent; PIE= excess relative risk due to mediation only or pure indirect effect; TE= Total excess relative risk; UK=United Kingdom. See supplementary Table 5 for protein abbreviations. Other Protein abbreviations are found at https://www.ncbi.nlm.nih.gov/gene/.

FIGURE 3. Four-way decomposition of the association between poor cardiovascular health as measured by LE8 total score (z-scored, reverse coded) and incidence of all-cause dementia by the selected plasma proteomic biomarkers (k=144): UK biobank 2006–2021

(A) Heatmap for raw 4-way decomposition: PIE p>0.05 (no significant pure mediation), k=77 (B) Heatmap for raw 4-way decomposition: PIE p<0.05, PIE<0, TE0, TE>CDE (consistent mediation), k=11 Note: Mediation Effects could range from 0 to 0.5. Abbreviations: CDE= excess relative risk due to neither mediation nor interaction or controlled direct effect; INTMED= excess relative risk due to mediated interaction or mediated interaction; INTREF= excess relative risk due to interaction only or interaction referent; PIE= excess relative risk due to mediation only or pure indirect effect; TE= Total excess relative risk; UK=United Kingdom. See supplementary Table 5 for protein abbreviations. Other Protein abbreviations are found at https://www.ncbi.nlm.nih.gov/gene/.