Cerebrospinal Fluid Dynamics and Intrathecal Delivery

Introduction to Pulsatile CSF Dynamics and CSF–Parenchyma–Vasculature Coupling

Current research into CSF velocities, CNS motion, and even respiratory influences are advancing the study of CSF dynamics, but critical questions remain. What is the exact nature of the force coupling between the pulsatile blood circulation with CSF displacements ( )? The exact source location and type of force coupling between the expanding and contracting vasculature, brain movement, and induced CSF dynamics are uncertain. This could elucidate key insights for transport phenomena of agents administered to the CSF. An additional question is whether perivascular solute transport is facilitated by naturally occurring convection under physiological conditions, which would require insights into the coupling between CSF motion and vascular pulsations. Also, the observed rapid penetration of tracer molecules through perivascular spaces observed in infusion experiments may have a strong pulsatile origin. The answer is of significant interest for treatment therapies of the CNS, where it could be exploited for achieving more rapid drug dispersion. A diagram of the main anatomical structures in the CNS is provided in Fig. 67.1 .

In our present view, we see CSF motion in the fluid-filled spaces mainly induced by vascular expansion. Imaging studies have confirmed that the cardiac cycle imposes its pulsatile pattern on to the CSF ( ). CSF also flows from the cranial to the spinal SAS in systole, with flow reversal from the spinal subarachnoid spaces into the cranium in diastole. Pulsatile CSF oscillations are believed to be driven by systolic vascular dilatation followed by diastolic contraction.

During the normal cardiac cycle, 550–950 mL of blood are pumped into the head per minute through the internal carotid artery and two vertebral arteries which discharge into the Circle of Willis before distribution to major cerebral territories ( ). The systolic pressure rise is believed to inflate blood vessels, thus augmenting the cerebral blood volume during systole. Due to the rigid cranial vault, the vascular expansion of the main cerebral arteries forces CSF displacement. This hypothesis is known as the Monro–Kellie doctrine. Magnetic resonance imaging (MRI) evidence suggests that the total cerebral blood volume inflates and deflates in each cardiac cycle by approximately 1–2 mL, the same volumetric amount, as there is CSF exchange between the cranial and spinal SAS ( ).

In addition to pulsations of the main arteries running along the cortical surface, smaller penetrating arterioles or the capillary bed, both of which are embedded inside the cortical tissue, may distend. Volumetric dilatation of the microcirculatory structures would need to be transmitted to the surrounding parenchyma. Brain tissue dilation would in turn have to displace interstitial fluid (ISF) through the extracellular space (ECS) into the ventricular system or produce brain motion ( ).

An MRI technique deployed by permitted the quantification of ventricular wall motion. This source of motion could originate from vascular volume dilatation transmitted from the cortical surface through brain tissue, whose compression causes ventricular contraction. Alternatively, ventricular wall motion could arise from inside the ventricles by systolic expansion of the choroidal arteries, such that ventricular walls pulsate against the periventricular ependymal layer. Ventricular dilation due to choroid expansion was hypothesized in a theoretical model and measured with cine phase-contrast MRI (PC MRI) ( ), which is a technique triggered by the heartbeat and captures data over a full cardiac cycle.

In addition to arterial expansion, compression of the venous system under high pressure has been hypothesized, which would reduce the venous blood lumen by cross-sectional area deformation or even collapse of sections of the venous tree ( ). speculated that the venous system is compressible, especially in patients with high intracranial pressure (ICP). Diminished venous lumen induces feedback that lowers blood flow, owing to the hike in resistance ( ). The possible collapse of vertebral blood flow when ICP exceeds a threshold (typically 30 mmHg) and reduced blood flow occurring in benign cerebral hypertension supports the notion of a compressible or collapsible venous bed ( ).

Moreover, the caudal decrease in CSF flow amplitude along the spinal canal ( ) supports the hypothesis of compliant tissue boundaries confining the CSF-filled spaces. Observed CNS compliance could result from periodic volume displacement within spinal or cranial veins, or the expansion of the elastic boundaries of the spinal SAS.

Respiratory Influence on CSF Flow

studied the effect of cardiac pulsations and respiration in spinal CSF movement in healthy volunteers. They concluded that cardiac pulsations are the main driver of CSF flow, but respiration plays a dominant role in the thoracolumbar region ( ). A respiratory influence on the CSF oscillations in the aqueduct has been observed in many studies ( ). The study by Dreha-Kulaczewski used volunteers who practiced forced breathing and breath holding, and found high CSF flow during inspiration phases. , used two-dimensional (2D) PC MRI to apply a correlation mapping technique to CSF flow in the cranial space of seven volunteers. The underlying finding was that the spatial distribution was different between the two components, suggesting different originating locations and mechanisms for driving CSF; one such example was the high correlation of cardiac pulsation in the prepontine region and high correlation for respiration along the venous sinus ( ).

employed real-time PC MRI with in vivo subjects to study the relative impact of respiration and cardiac pulsations on CSF flow dynamics using an MR technique to detect signal frequencies not aligned with the cardiac output. In human volunteers they found deep controlled breathing forced CSF cephalad at the foramen magnum during inhalation and caudad during exhalation. CSF flow during natural breathing was more influenced by cardiac pulsations.

Growing interest in the respiratory component as a driving motor for CSF flow has elucidated that both respiration and cardiac pulsations influence CSF motion. The respiration influence has been identified as distinct from the cardiac phase, and the respiratory influence has been shown to be strongest in the thoracolumbar region. Forced respiration was shown to override cardiac pulsations, but the forced exercise does not represent the natural relationship between cardiac pulsations and baseline breathing. The question of relative strength of influence on CSF flow for cardiac versus respiratory driving forces remains undetermined. Further studies are needed to disentangle the relative effects between breathing and cardiac cycle, as well as spinal and cranial compliance. Patient-specific factors such as spinal and cranial compliance, CSF volumes, and heart rate and respiration frequency variation may prohibit a generalized scheme for attributing CSF dynamics to individual driving components. The open question regarding the driving motor of CSF flow lends weight to the argument to consider each subject individually.

Geometry-Induced Flow Phenomena (Nerve Roots and Trabeculae)

Nerve roots connecting the CNS with the peripheral nerves extend between the spinal pia mater and the outer dura mater crossing the spinal CSF-filled space. The protruding nerve roots constitute geometric obstructions to unhindered CSF flow, inducing vigorous steering in the spinal CSF with eddies and recirculation zones. Microanatomical features causing complex flows also include ligaments, spinal–arachnoid midline septa, trabeculae, and meningeal layers, which were carefully characterized by .

Solute dispersion in an annular spine segment was studied by . Microanatomical features increased solute dispersion up to 10-fold when compared to a simulation on a model without nerve roots and trabeculae. Biodistribution of an IT-administered drug was studied by Hsu et al. in an idealized axisymmetric spinal and cranial SAS ( ). The authors report that pulse amplitude and frequency were key factors affecting the speed and spread of drug distribution. Tangen et al. predicted drug distribution in a subject-specific three-dimensional (3D) model of the entire CNS, confirming that anatomical features in the spine enhance mixing and the rapid spread of the drug ( ). The computational analyses revealed that nerve roots and trabeculae create vortices, which break the laminar flow profiles in the pulsating CSF of the spine and thus increase the rate of drug spread. In effect, convective drug transport is rapid despite a low Reynolds number and slow drug diffusivity. Our characterization of the geometry-induced flow regimes, also known as chaotic advection ( ), with associated micromixing patterns has significant implications for local anesthesia and chronic pain management.

Cilia-Induced CSF Flow in the Cerebral Ventricles

Recently ependymal cilia lining the cerebral ventricles were shown to influence near-wall CSF flow in mice ( ). A study on the third ventricle of the mouse brain elucidated a transport network driven by organized cilia motion. Additionally, these cilia were able to induce complex flow patterns and change flow direction. The overall contribution of these near-wall flow effects is an ongoing area of research, but they could have a potential role in substance distribution within the ventricles ( ).

Acquisition of CNS Anatomy

MRI is a standard technique to measure CSF flow velocities and acquire anatomical structures. CSF exhibits dark signals in T1-weighted images, which are commonly used in image segmentation for reconstructing models of cerebral structures, including the CSF space, gray matter, and white matter. Multiple groups have proposed automatic or semiautomatic image-processing algorithms for segmentation of the CSF space ( ). T2-weighted images capture CSF as a bright signal and are widely used for clinical diagnosis of CSF-related diseases, including hydrocephalus and aqueduct stenosis ( ). CSF flow measurements require a different imaging protocol, briefly described next.

CSF Flow Measurements