Associations of Blood Cardiovascular Biomarkers With Brain Free Water and Its Relationship to Cognitive Decline: A Diffusion-MRI Study

Abstract

Background and objectives: There is an increasing awareness of the "Heart-Brain Connection," whereby cardiovascular function is connected with cognition. Diffusion-MRI studies reported higher brain free water (FW) was associated with cerebrovascular disease (CeVD) and cognitive impairment. In this study, we investigated whether higher brain FW was related to blood cardiovascular biomarkers and whether FW mediated the associations between blood biomarkers and cognition.

Methods: Participants recruited from 2 Singapore memory clinics between 2010 and 2015 underwent collection of blood samples and neuroimaging at baseline and longitudinal neuropsychological assessments up to 5 years. We examined the associations of blood cardiovascular biomarkers (high-sensitivity cardiac troponin-T [hs-cTnT], N-terminal pro-hormone B-type natriuretic peptide [NT-proBNP], and growth/differentiation factor 15 [GDF-15]) with brain white matter (WM) and cortical gray matter (GM) FW derived from diffusion MRI using whole brain voxel-wise general linear regression. We then assessed the relationships among baseline blood biomarkers, brain FW, and cognitive decline using path models.

Results: A total of 308 older adults (76 with no cognitive impairment, 134 with cognitive impairment no dementia, and 98 with Alzheimer disease dementia and vascular dementia; mean [SD] age: 72.1 [8.3]) were included. We found that blood cardiovascular biomarkers were associated with higher FW in widespread WM regions and in specific GM networks including the default mode, executive control, and somatomotor networks at baseline (p < 0.01, family-wise error corrected). Baseline FW in widespread WM and network-specific GM fully mediated the associations of blood biomarkers with longitudinal cognitive decline over 5 years. Specifically, in GM, higher FW in the default mode network mediated the relationship with memory decline (hs-cTnT: β = -0.115, SE = 0.034, p = 0.001; NT-proBNP: β = -0.154, SE = 0.046, p = 0.001; GDF-15: β = -0.073, SE = 0.027, p = 0.006); by contrast, higher FW in the executive control network was responsible for executive function decline (hs-cTnT: β = -0.126, SE = 0.039, p = 0.001; NT-proBNP: β = -0.110, SE = 0.038, p = 0.004; GDF-15: β = -0.117, SE = 0.035, p = 0.001). Similar full mediation effects of brain FW were also identified for baseline cognition.

Discussion: Results suggested a role of brain FW in linking cardiovascular dysfunction to cognitive decline. These findings provide new evidence for brain-heart interactions, paving the way for prediction and monitoring of domain-specific cognitive trajectory.

Figure 1. Study Design Schematic

A total of 308 participants with either NCI, CIND or dementia were studied. GLMs were performed to identify region-specific associations between brain FW and the 3 blood cardiovascular biomarkers within the CIND + dementia group or NCI at baseline (step 1). Path analyses were used to evaluate whether and how FW in the gray and white matter mediated the effects of higher blood marker levels on baseline global cognitive deficits and longitudinal decline (step 2). Furthermore, the influences of network-specific GM FW on the association of higher blood marker levels with individual cognitive domains were also evaluated. CIND = cognitive impairment with no dementia; FW = free-water; GLM = general linear model; GM = gray matter; NCI = no cognitive impairment; WM = white matter.

Figure 2. Higher White Matter FW Correlated With Circulating Cardiovascular Marker Levels

The whole-brain voxel-wise linear regression indicated that higher FW values in widespread brain white matter regions were associated with increased levels of (A) hs-cTNT, (B) NTpro-BNP, and (C) GDF-15. Results are TFCE enhanced, reported at p < 0.01, FWE corrected. FW = free water; FWE = family-wise error; GDF-15 = growth/differentiation factor 15; hs-cTnT = high-sensitivity cardiac troponin-T; NT-proBNP = N-terminal pro hormone B-type natriuretic peptide; TFCE = threshold-free cluster enhancement.

Figure 3. Higher Gray Matter FW Correlated With Circulating Cardiovascular Marker Levels

The whole-brain vortex-wise linear regression indicated that higher FW values in middle frontal and temporal lobes and cingulate regions were associated with increased levels of (A) hs-cTNT, (B) NTpro-BNP, and (C) GDF-15. (D) Yeo's functional intrinsic networks parcellation. Results are reported at p < 0.01, FWE corrected. DAN = dorsal attention network; DMN = default mode network; ECN = executive control network; FW = free water; GDF-15 = growth/differentiation factor 15; hs-cTnT = high-sensitivity cardiac troponin-T; NT-proBNP = N-terminal pro hormone B-type natriuretic peptide; SMN = somatomotor network; SN/VA = ventral attention network.

Figure 4. Effects of Circulating Cardiovascular Biomarker Levels on Global Cognition Through Mediators

Schematic diagram of the path analyses for hs-cTNT. hs-cTNT was entered as a predictor in each model. The mean WM FW and mean GM FW were added as mediators. (A) Global cognition baseline impairment and (B) rate of decline over time (5 years) were treated as the outcome. Numbers on the paths indicate standardized coefficients that were statistically significant. FW = free water; GM = gray matter; hs-cTnT = high-sensitivity cardiac troponin-T; WM = white matter.

Figure 5. Effects of Circulating Cardiovascular Biomarker Levels on Longitudinal Executive Function and Memory Decline Through Mediators

Schematic diagram of the path analyses. hs-cTNT was inputted as a predictor. The mean WM FW and mean GM FW derived from 7 Yeo's intrinsic networks were added as mediators (see parcellations with colored boundary in the right bottom). Rates of changes in executive and memory domains over time (5 years) were treated as outcomes. Numbers on the paths indicate standardized coefficients that were statistically significant. DAN = dorsal attention network; DMN = default mode network; ECN = executive control network; FW = free water; GM = gray matter; hs-cTnT = high-sensitivity cardiac troponin-T; SMN = somatomotor network; SN/VA = ventral attention network; WM = white matter.