Umeå University's logo

Background: Periarterial cerebrospinal fluid (CSF) flow has been hypothesized to contribute to brain waste clearance but is poorly understood. Animal studies suggest arterial pulsatility drives perivascular CSF in the direction of cerebral arterial blood flow (CBF), but human validation has relied on MRI approaches that do not inform on flow-directionality. Here, we use a high-performance gradient system enabling low velocity encoding (Venc) 4D flow MRI, to characterize cardiac-driven periarterial CSF flow and CBF-to-CSF flow coupling.

Methods: Healthy participants (N = 10) underwent high resolution (0.8 mm isotropic) 4D flow MRI of blood (Venc = 120 cm/s) and CSF (Venc = 1.0 cm/s) on a 3.0T head-only system (MAGNUS, GE Healthcare; Gmax = 300mT/m, Smax = 750T/m/s). Images were reconstructed using local low rank (LLR) constrained parallel imaging and background field corrected using iterative, complex domain fitting. Luminal blood and associated periarterial CSF waveforms were extracted along the left and right anterior (ACA A1), middle (MCA M1 and M2), and posterior (PCA P2) cerebral arteries using a centerline approach, and characterized individually by amplitude and stroke volume and jointly by coupling coefficients from maximum cross-correlation () and time-lags.

Results: For low Venc 4D flow MRI, MAGNUS dramatically reduced the echo time and temporal resolution compared to whole-body systems. CBF and CSF measurements were successful in 61/80 locations (up to 8 per participant) with 19 measurements excluded due to velocity aliasing and/or poor local quality of the flow data. Inverse (anti-correlated) CBF-to-CSF coupling () was observed for most segments (56/61), with strong coupling observed for all vessels, including M1 (-0.85 ± 0.06), A1 (-0.80 ± 0.12), P2 (-0.79 ± 0.08), and M2 (-0.78 ± 0.08). Further, CBF preceded CSF for most (43/56) segments, with short CBF-to-CSF lags in A1 (5.30 ± 64 ms) and P2 (4.13 ± 63 ms), higher in M1 (43 ± 39 ms), and highest in M2 (115 ± 39 ms). CBF and CSF flow metrics were also correlated in terms of flow rate amplitude (r = 0.40, p = 0.015) and stroke volume (r = 0.56, p < 0.001).

Conclusions: High-performance gradient systems facilitate 4D flow imaging of very slow CSF. Joint analysis of CBF and periarterial CSF allowed assessment of CBF-to-CSF dynamics coupling. For most vessels, an inverse coupling and a positive time-lag was found from CBF to periarterial CSF, suggesting that the systolic arterial expansion drives CSF backwards and inwards again during diastolic relaxation. The proposed approach can be used to improve our understanding of CBF and CSF dynamics in aging and dementia. Clinical trial number: Not applicable.

Background: White matter lesions (WML) and dilated perivascular spaces (PVS) are features of small vessel disease (SVD), commonly observed in aging and dementia, with unknown pathophysiology. Human studies have documented contrast accumulation within and in proximity of SVD-lesions. However, whether such observations mainly reflect excessive blood-brain barrier (BBB) leakage, or altered microvascular density in the investigated regions, remains unclear.

Methods: To evaluate the roles of BBB leakage and vascular density in aging and SVD, dynamic contrast enhanced (DCE) MRI was used to estimate the permeability-surface area product (PS) and fractional plasma volume () in normal-appearing brain tissue and in proximity of and within WML and PVS in a population-based cohort (N = 56; 34/22 m/f; age 64 to 84 years). Analysis of variance (ANOVA) was used to assess regional differences in PS and and analysis of covariance (ANCOVA) was used to assess regional differences in PS with and vascular risk as covariates.

Results: Pronounced increases in PS and were observed from normal-appearing white matter (NAWM) to WML peripheries to WMLs. Similar PS and increases were observed from basal ganglia (BG) to BG-PVS. Further, PS in NAWM and white matter (WM) PVS were found to increase with cortex-to-ventricular depth. However, ANCOVA models with as a covariate showed that variance in PS was mainly explained by vp (η2=0.17 to η2=0.35; all p < 10− 3), whereas the effect of region was only borderline-significant when comparing NAWM, WML peripheries and WML (p = 0.03) and non-significant for the other comparisons (p > 0.29).

Conclusions: Our findings support the notion that contrast leakage across the BBB accumulates within and in proximity of SVD-related lesions. However, high contrast accumulation may mainly reflect high vascularization, and to a lesser degree than previously recognized BBB dysfunction.

Neurofluid dynamics are crucial for maintaining brain homeostasis and facilitating the clearance of brain metabolites through the coupling of arterial and venous blood with cerebrospinal fluid (CSF). Two-dimensional phase contrast (PC) magnetic resonance imaging (MRI) is frequently used to study neurofluids; however, separate examinations are typically required for assessing blood and CSF flow, which can confound analyses due to asynchronous physiological measurements. To enable simultaneous assessment of neurofluid dynamics, we describe and evaluate a 2D PC MRI approach in human participant experiments. An interleaved multi-point velocity encoding scheme was integrated into a 2D golden angle spiral PC MRI scan to facilitate synchronous characterization of neurofluids. Two multi-point schemes, including interleaved dual-venc (DV) and triple-venc (MV) scans, were evaluated and compared with standard asynchronous single-venc (SV) scans. Data and repeated scans were collected on a clinical 3.0T scanner at the level of the C1/C2 vertebrae in 10 human participants. From cardiac-resolved images, the relationship between net blood flow and CSF flow pulsatile volume change was characterized using regression modeling. Temporal lags between cardiac-driven arterial blood (vertebral arteries (VAs) and internal carotid arteries (ICAs)) and spinal canal (SC) CSF were estimated with cross-correlation. SV, DV, and MV flow mean, range, and volume changes were studied and compared using linear mixed effect models, intraclass correlation coefficients, Bland–Altman, and Pearson correlations. A strong relationship was measured between net blood flow and CSF flow pulsatile volume change from SV (R2 = 0.71, P = 0.002), DV (R2 = 0.70, P = 0.003), and MV (R2 = 0.78, P < 0.001) scans. SC-VAs temporal lags were statistically longer than SC-ICAs lags across all scans (P < 0.001 for SV, DV, and MV). Bland–Altman analyses and repeatability coefficients indicated that DV and MV scans had the highest repeatability. MV scans generally underestimated SC CSF flow markers relative to SV and DV scans. A more pronounced flow offset in venous measures was identified between SV scans and the DV and MV scans. In conclusion, this study introduced a method for simultaneous imaging of cranio-spinal arterial, venous, and CSF flow, enabling the synchronous assessment of neurofluid dynamics. The results indicated that interleaved DV and MV scans could improve the evaluation of neurofluid coupling compared with asynchronous SV scans.

Background: Neurofluid flow dynamics are frequently studied from asynchronous blood and CSF flow measurements from real-time imaging using separate phase contrast (PC) MRI scans. Asynchronous measures can be influenced by changes in heart rate, respiration, and other physiological processes, obfuscating neurofluids assessment. Here we present an approach for synchronous measures of neurofluids using simultaneous real-time 2D PC MRI and investigated the effects of different breathing patterns on synchronous and asynchronous blood and CSF flow in a group of healthy participants.

Methods: Interleaved dual-velocity encoding 2D PC MRI with retrospective real-time reconstruction was utilized for synchronous neurofluid measures during free breathing, paced breathing and breath holds. Data were collected on a clinical 3.0T scanner at the level of C1/C2 vertebrae in 10 participants. From real-time images, flow rates repeated measures, and cardiac and respiratory flow power were assessed using Bland-Altman, power spectral analyses, and breathing pattern group differences. Neurofluids coupling from cross-correlation between arterial and venous blood and CSF flow signals were quantified from synchronous and asynchronous measures. Real-time images were re-binned to the cardiac cycle and compared to high-temporal resolution cardiac-resolved images in terms of flow rate, pulsatility index, and stroke volume.

Results: Flow repeatability was highest in free breathing scans and in arteries and spinal canal compared to veins from Bland-Altman and repeatability coefficients. Significant differences were measured in cardiac and respiratory flow power across breathing patterns in various vessel segments and spinal canal (P ≤ 0.006). Synchronous blood and CSF cardiac coupling were significantly higher than asynchronous results in all vessels (P = 0.002). For example, free breathing synchronous cardiac couplings ranged from [0.81, 0.93], compared to asynchronous range [0.49, 0.53]. Synchronous cardiac coupling showed significant differences across breathing patterns in most vessels (P = 0.002). Comparison between real-time cardiac re-binned images and high-temporal resolution cardiac-resolved images showed high correlations for flow rate and spinal canal stroke volume (ρ ≥ 0.95) and lower for pulsatility index (ρ = [0.45, 0.88]).

Conclusions: Breathing patterns induced significant responses across neurofluids including in flow rates, flow power, and coupling parameters. Higher cross-correlation among synchronous measures support benefits over asynchronous measures for neurofluids coupling characterization.

Blood-brain barrier (BBB) disruption may contribute to cognitive decline, but questions remain whether this association is more pronounced for certain brain regions, such as the hippocampus, or represents a whole-brain mechanism. Further, whether human BBB leakage is triggered by excessive vascular pulsatility, as suggested by animal studies, remains unknown. In a prospective cohort (N = 50; 68-84 years), we used contrast-enhanced MRI to estimate the permeability-surface area product (PS) and fractional plasma volume ( formula presented ), and 4D flow MRI to assess cerebral arterial pulsatility. Cognition was assessed by the Montreal Cognitive Assessment (MoCA) score. We hypothesized that high PS would be associated with high arterial pulsatility, and that links to cognition would be specific to hippocampal PS. For 15 brain regions, PS ranged from 0.38 to 0.85 (·10-3 min-1) and formula presented from 0.79 to 1.78%. Cognition was related to PS (·10-3 min-1) in hippocampus (β = - 2.9; p = 0.006), basal ganglia (β = - 2.3; p = 0.04), white matter (β = - 2.6; p = 0.04), whole-brain (β = - 2.7; p = 0.04) and borderline-related for cortex (β = - 2.7; p = 0.076). Pulsatility was unrelated to PS for all regions (p > 0.19). Our findings suggest PS-cognition links mainly reflect a whole-brain phenomenon with only slightly more pronounced links for the hippocampus, and provide no evidence of excessive pulsatility as a trigger of BBB disruption.

Background: Cerebrospinal fluid (CSF) dynamics are increasingly studied in aging and neurological disorders. Models of CSF-mediated waste clearance suggest that altered CSF dynamics could play a role in the accumulation of toxic waste in the CNS, with implications for Alzheimer’s disease and other proteinopathies. Therefore, approaches that enable quantitative and volumetric assessment of CSF flow velocities could be of value. In this study we demonstrate the feasibility of 4D flow MRI for simultaneous assessment of CSF dynamics throughout the ventricular system, and evaluate associations to arterial pulsatility, ventricular volumes, and age.

Methods: In a cognitively unimpaired cohort (N = 43; age 41–83 years), cardiac-resolved 4D flow MRI CSF velocities were obtained in the lateral ventricles (LV), foramens of Monro, third and fourth ventricles (V3 and V4), the cerebral aqueduct (CA) and the spinal canal (SC), using a velocity encoding (venc) of 5 cm/s. Cerebral blood flow pulsatility was also assessed with 4D flow (venc = 80 cm/s), and CSF volumes were obtained from T1- and T2-weighted MRI. Multiple linear regression was used to assess effects of age, ventricular volumes, and arterial pulsatility on CSF velocities.

Results: Cardiac-driven CSF dynamics were observed in all CSF spaces, with region-averaged velocity range and root-mean-square (RMS) velocity encompassing from very low in the LVs (RMS 0.25 ± 0.08; range 0.85 ± 0.28 mm/s) to relatively high in the CA (RMS 6.29 ± 2.87; range 18.6 ± 15.2 mm/s). In the regression models, CSF velocity was significantly related to age in 5/6 regions, to CSF space volume in 2/3 regions, and to arterial pulsatility in 3/6 regions. Group-averaged waveforms indicated distinct CSF flow propagation delays throughout CSF spaces, particularly between the SC and LVs.

Conclusions: Our findings show that 4D flow MRI enables assessment of CSF dynamics throughout the ventricular system, and captures independent effects of age, CSF space morphology, and arterial pulsatility on CSF motion.

Neurological disorders can manifest with altered neurofluid dynamics in different compartments of the central nervous system. These include alterations in cerebral blood flow, cerebrospinal fluid (CSF) flow, and tissue biomechanics. Noninvasive quantitative assessment of neurofluid flow and tissue motion is feasible with phase contrast magnetic resonance imaging (PC MRI). While two-dimensional (2D) PC MRI is routinely utilized in research and clinical settings to assess flow dynamics through a single imaging slice, comprehensive neurofluid dynamic assessment can be limited or impractical. Recently, four-dimensional (4D) flow MRI (or time-resolved three-dimensional PC with three-directional velocity encoding) has emerged as a powerful extension of 2D PC, allowing for large volumetric coverage of fluid velocities at high spatiotemporal resolution within clinically reasonable scan times. Yet, most 4D flow studies have focused on blood flow imaging. Characterizing CSF flow dynamics with 4D flow (i.e., 4D CSF flow) is of high interest to understand normal brain and spine physiology, but also to study neurological disorders such as dysfunctional brain metabolite waste clearance, where CSF dynamics appear to play an important role. However, 4D CSF flow imaging is challenged by the long T1 time of CSF and slower velocities compared with blood flow, which can result in longer scan times from low flip angles and extended motion-sensitive gradients, hindering clinical adoption. In this work, we review the state of 4D CSF flow MRI including challenges, novel solutions from current research and ongoing needs, examples of clinical and research applications, and discuss an outlook on the future of 4D CSF flow.

Objective: High cerebral arterial pulsatility index (PI), white matter lesions (WMLs), enlarged perivascular spaces (PVSs), and lacunar infarcts are common findings in the elderly population, and considered indicators of small vessel disease (SVD). Here, we investigate the potential temporal ordering among these variables, with emphasis on determining whether high PI is an early or delayed manifestation of SVD.

Methods: In a population-based cohort, 4D flow MRI data for cerebral arterial pulsatility was collected for 159 participants at baseline (age 64–68), and for 122 participants at follow-up 5 years later. Structural MRI was used for WML and PVS segmentation, and lacune identification. Linear mixed-effects (LME) models were used to model longitudinal changes testing for pairwise associations, and latent change score (LCS) models to model multiple relationships among variables simultaneously.

Results: Longitudinal 5-year increases were found for WML, PVS, and PI. Cerebral arterial PI at baseline did not predict changes in WML or PVS volume. However, WML and PVS volume at baseline predicted 5-year increases in PI. This was shown for PI increases in relation to baseline WML and PVS volumes using LME models (R (Formula presented.) 0.24; p < 0.02 and R (Formula presented.) 0.23; p < 0.03, respectively) and LCS models ((Formula presented.) = 0.28; p = 0.015 and (Formula presented.) = 0.28; p = 0.009, respectively). Lacunes at baseline were unrelated to PI.

Interpretation: In healthy older adults, indicators of SVD are related in a lead–lag fashion, in which the expression of WML and PVS precedes increases in cerebral arterial PI. Hence, we propose that elevated PI is a relatively late manifestation, rather than a risk factor, for cerebral SVD. 

4D flow magnetic resonance imaging (MRI) is increasingly recognizedas a versatile tool to assess arterial and venous hemodynamics. Cerebral arterial pulsatility is typically assessed by analyzing flow waveforms over the cardiac cycle, where flow amplitude is a function of cardiac output, central arterial stiffness, and cerebrovascular resistance and compliance. Excessive pulsatility may propagate to the cerebral microcirculation, and constitute a harmful mechanism for the brain. Indeed, imaging studies have linked arterial pulsatility to hippocampus volume, cerebral small vessel disease (SVD), and Alzheimer’s disease (AD). In animal models, elevated pulsatility leads to blood-brain barrier (BBB) leakage, capillary loss, and cognitive decline. However, associations to cerebrovascular lesions and brain function in the spectrum of normal aging are less investigated. Further, previous 4D flow studies have mainly assessed pulsatility in relatively large cerebral arteries. When exploring links to microvascular damage and brain function, more distal measurements, closer to the microcirculation, are desired. 

This thesis aimed to develop 4D flow MRI post-processing methods to obtain pulsatile waveforms in small, distal cerebral arteries with noisy velocity data and a complex vascular anatomy, and to evaluate pulsatility (primarily assessed by the pulsatility index) in relation to aging, brain function, and other imaging biomarkers of cerebrovascular damage, with particular dedication towards the hippocampus and cerebral SVD. 

To assess pulsatility in distal cerebral arteries, a post-processing method that automatically samples waveforms from numerous small arteries, to obtain a whole-brain representation of the distal arterial waveform, was developed (Paper I). We demonstrated the importance of averaging flow waveforms along multiple vessel segments to avoid overestimations in the pulsatility index, showed agreement with reference methods, and linked distal arterial pulsatility to age. 

To explore links to hippocampal function, we evaluated pulsatility in relation to cognition, hemodynamic low-frequency oscillations (LFOs), perfusion, and hippocampus volume (Paper II). We found that higher pulsatility was linked to worse hippocampus-sensitive episodic memory, weaker hippocampal LFOs, and lower whole-brain perfusion. These findings aligned with models suggesting that hippocampal microvessels could be particularly susceptible to pulsatile stress.

To inform on SVD pathophysiology, we evaluated 5-year associations among pulsatility, white matter lesions (WMLs) and perivascular space (PVS) dilation, using mixed models, factor analysis, and change score models (Paper III). Lead-lag analyses indicated that, while pulsatility at baseline could not predict WML nor PVS progression, WML and PVS volumes at baseline predicted 5-year pulsatility-increases. These findings indicate that individuals with a higher load of cerebrovascular damage are more prone to see increased pulsatility over time, and suggest that high pulsatility is a manifestation, rather a risk factor, for cerebral SVD.   

To shed light on the potential role of BBB leakage in aging and SVD, we used dynamic contrast enhanced (DCE) MRI and intravenous gadolinium injections to quantify BBB permeability (Paper IV). We found stepwise increases in permeability from healthy white matter to WMLs, supporting that BBB leakages are implicated in SVD. However, hippocampal BBB permeability was unrelated to age, indicating that this capillary property is maintained in aging. Finally, arterial pulsatility was unrelated to BBB permeability in WMLs and in the hippocampus, providing no evidence of excessive pulsatility as a trigger of BBB leakage. 

In conclusion, distal arterial pulsatility measurements are reliable when averaging 4D flow waveforms over a large number of vessels. Pulsatility increases with age, and individuals with more cerebrovascular lesions are prone to see larger increases over time. Pulsatility is negatively related to perfusion and hippocampal function. However, the temporal dynamics among the SVD biomarkers, and the absence of pulsatility–permeability associations, challenge the concept of excessive pulsatility as a trigger of microvascular damage. Future studies are needed to understand whether altered cerebral hemodynamics play a causal role in cognitive decline and dementia. Meanwhile, 4D flow hemodynamic parameters could be useful as biomarkers related to vessel properties and cerebrovascular health. 

Cognitive functions are well-preserved for some older individuals, but the underlying brain mechanisms remain disputed. Here, 5-year longitudinal 3-back in-scanner and offline data classified individuals in a healthy older sample (baseline age = 64–68 years) into having stable or declining working-memory (WM). Consistent with a vital role of the prefrontal cortex (PFC), WM stability or decline was related to maintained or reduced longitudinal PFC functional responses. Subsequent analyses of imaging markers of general brain maintenance revealed higher levels in the stable WM group on measures of neurotransmission and vascular health. Also, categorical and continuous analyses showed that rate of WM decline was related to global (ventricles) and local (hippocampus) measures of neuronal integrity. Thus, our findings support a role of the PFC as well as general brain maintenance in explaining heterogeneity in longitudinal WM trajectories in aging.