CNS wide simulation of flow resistance and drug transport due to spinal microanatomy
Introduction
Despite precise MR observations of cerebrospinal fluid (CSF) velocities (Laitinen 1968), brain motion (Zhu et al., 2006) and ventricular wall dilatations (Enzmann and Pelc, 1992), critical questions remain about CSF dynamics in the CNS: What is the exact nature of the force coupling between the pulsatile blood circulation and CSF displacements? (Marmarou et al., 1975, Marmarou et al., 1978) How is the pulsatile expansion and contraction of the cerebrovascular bed transmitted to the CSF and the brain? The source of the volumetric dilations that actually induce CSF motion remain uncertain, even though major cerebral arteries, the oscillatory displacement by the parenchyma or choroid plexus expansion are likely involved. Moreover, the caudal decrease in CSF flow amplitude along the spinal canal (Alperin et al., 2005, Tain et al., 2011) supports the notion of compliant leptomeningeal boundaries lining the fluid-filled spinal subarachnoid space (SAS). Hence, the observed CNS compliance could result from cyclic displacement of blood inside large spinal veins, or the expansion of CSF-filled spaces against elastically deformable epidural tissues. Nevertheless, the biomechanics of deformations of the entire spinal compartment remain unclear.
Subject-specific CSF flow measurements may be significant for improving noninvasive diagnostics of pathological conditions. For example, the amplitude ratio between the aqueduct and prepontine CSF flows were found elevated in hydrocephalus patients compared to normal subjects (Zhu et al., 2006). Because in vivo MR measurements have limited image resolution and practical scan durations restrict the number of observed regions of interest, many researchers have attempted to address fundamental questions regarding CNS with computational fluid dynamics (CFD) techniques dynamics especially under pathological conditions.
Early CFD models used small sections of the CNS with idealized geometries (Gupta and Poulikakos, 2008, Kurtcuoglu et al., 2005, Kurtcuoglu et al., 2007, Linge et al., 2010, Loth et al., 2000). Other approaches used medical image reconstruction to represent subjects-specific anatomical spaces to enhance the model fidelity (Gupta et al., 2010, Linninger et al., 2007, Roldan et al., 2009, Rutkowska et al., 2012, Somayaji et al., 2008, Sweetman and Linninger, 2011, Sweetman et al., 2011). Haller and Low (1971) showed that spinal microanatomy induces complex mixing eddies. Stockman studied the impact of nerve roots and trabeculae on convective flow with Lattice Boltzman simulations in idealized cylindrical spine models (Stockman, 2005, Stockman, 2007). Gupta presented an analytical solution to oscillating fluid in a representative elliptical annulus representing a spinal segment directly inferior to the cisterna magna (Gupta et al., 2009). Based on a simulated porous model of the cerebromeduallary cistern and spinal SAS, they conclude that subresolution microstructure density and radius can triple the pressure drop in CSF flow (Gupta et al., 2009, Gupta et al., 2010).
Several groups investigated CSF flow patterns in Chiari patients relative to normal subjects (Martin et al., 2013, Roldan et al., 2009, Rutkowska et al., 2012). Yiallourou et al. (2012) compared flow measurements in normal subjects and patients suffering from Chiari malformation with a CFD model of a cervical segment of the spine. During the review process of this paper, an updated report to the earlier CFD simulation paper of the same group including artificial nerve roots appeared (Heidari Pahlavian et al., 2014). Yet none of these studies performed CFD simulations on the entire CNS and included microanatomical trabeculae in the spinal SAS.
The goal of this paper is to address several critical questions regarding the spatiotemporal coupling of the cranio-spinal CSF compartments by performing CFD simulation on an entire CNS model of a normal subject. This study tests the hypothesis that microanatomical structures generate significant vortex phenomena in spinal CSF flow which increase flow resistance, raise pressure drop, and induce complex mixing patterns responsible for the rapid biodispersion of moieties administered to the spinal SAS. To account for the influence of detailed spinal anatomical structures, nerve roots were reconstructed from medical images; arachnoid trabeculae below the imaging threshold were artificially incorporated to the subject-specific computational mesh. This paper constitutes a first step towards a complete in silico model of CNS dynamics with full coupling between CSF, blood and brain. To augment the limited scope of previous studies, it is critical to account for the entire CNS including the cranial and spinal SAS with their microscopic aspects. Modeling the complete CNS eliminates the burden of introducing uncertain assumptions about internal boundary conditions. Therefore, a parametric study to quantify the influence of microanatomical aspects on CSF dynamics and flow resistance and drug transport was conducted.
Section snippets
CSF spaces, in vivo velocity measurements and CFD simulations
Methods used are summarized here in brief, while details are presented in the Supplementary Material. A subject-specific computational model of the cranial and spinal CSF-filled spaces including nerve roots was reconstructed from medical images, up to detail limited by resolution of the dataset using MIMICS image processing software (Appendix 1.2). CSF flow measurements were acquired from the 29 year old volunteer with CINE MRI at three planes in the spinal SAS (Appendix 1.1).
The cranial pial
Analysis of CSF flow fields in the full SAS model
The simulated CSF flow velocities are visualized as contour plots on axial planes of interest for a full cardiac cycle, in Fig. 2. Systole lasts about 25–35% of the cardiac cycle with a sharp forceful peak in the craniocaudal direction, this phase is color coded in red. Diastole covers the remainder of the cardiac cycle with a slower, more even flow in the caudocranial direction, color coded in blue. Regions of low velocities, ||v||<0.1 cm s−1, are coded in green to highlight temporary stagnation
Discussion
This computational study aimed at predicting the CSF dynamics in a subject-specific model of the entire CNS including microanatomical aspects. The massive CFD computations assessed the effect of microanatomy on CSF flow and flow resistance. Simulations found the caudal flow phase to last 20–30% of the cardiac cycle, which agreed with MRI measurements. The cisterna magna had a stroke volume of 0.76 mL with caudal decrease in amplitudes.
Pressure pulsatility along the spinal canal was
Conclusions and limitations
The dynamic 3D model presented above is a first attempt to predict CSF circulation in the entire CNS of an individual subject. The timing of the pressure and fluid velocity fields was carefully quantified and encoded in a comprehensive Fourier Map. The microcirculatory flow pattern introduced by nerve roots and arachnoid trabeculae were characterized with the use of massively parallel supercomputer simulations.
The actual functional role of trabeculae has not been established, but our results
Conflict of interest
None of the authors has a conflict of interest regarding this work.
Acknowledgments
The authors would like to gratefully acknowledge partial support from the National Science Foundation Grant CBET-0756154. This work used the Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by National Science Foundation Grant number OCI-1053575. Specifically, it used the Blacklight system at the Pittsburgh Supercomputing Center (PSC). We would like to thank C. Wivel from Materialize Inc. for providing a research license for Mimics image reconstruction software.
References (40)
- et al.
Compensated hyperosmolarity of cerebrospinal fluid and the development of hydrocephalus
Neuroscience
(2013) - et al.
Computational investigation of subject-specific cerebrospinal fluid flow in the third ventricle and aqueduct of Sylvius
J. Biomech.
(2007) - et al.
Mixing and dispersion in a baffled tube for steady laminar and pulsatile flow
Chem. Eng. Sci.
(1991) - et al.
Development of hydrocephalus and classical hypothesis of cerebrospinal fluid hydrodynamics: facts and illusions
Prog. Neurobiol.
(2011) - et al.
Gap instability of laminar flows in eccentric annular channels
Nucl. Eng. Des.
(2011) - et al.
Systematic design of drug delivery therapies
Comput. Chem. Eng.
(2008) - et al.
Three-dimensional computational prediction of cerebrospinal fluid flow in the human brain
Comput. Biol. Med.
(2011) Microscopic morphology and histology of the human meninges
Morphol. Bull. Assoc. Anat.
(2005)- et al.
MRI study of cerebral blood flow and CSF flow dynamics in an upright posture: the effect of posture on the intracranial compliance and pressure
Acta Neurochir. Suppl.
(2005) - et al.
Magnetic resonance imaging-based measurements of cerebrospinal fluid and blood flow as indicators of intracranial compliance in patients with Chiari malformation
J. Neurosurg.
(2005)
Scanning electron microscopy of the subarachnoid space in the dog. I. Spinal cord levels
J. Comp. Neurol.
Brain motion: measurement with phase-contrast MR imaging
Radiology
Analytical solution for pulsatile viscous flow in a straight elliptic annulus and application to the motion of the cerebrospinal fluid
Phys. Fluids
Three-dimensional computational modeling of subject-specific cerebrospinal fluid flow in the subarachnoid space
J. Biomech. Eng.
Cerebrospinal fluid dynamics in the human cranial subarachnoid space: an overlooked mediator of cerebral disease. I. Computational model
J. R. Soc. Interface
The fine structure of the peripheral nerve root sheath in the subarachnoid space in the rat and other laboratory animals
Am. J. Anat.
The Impact of spinal cord nerve roots and denticulate ligaments on cerebrospinal fluid dynamics in the cervical spine
PLoS ONE
The effect of pulsatile flow on intrathecal drug delivery in the spinal canal
Ann. Biomed. Eng.
The frequency and magnitude of cerebrospinal fluid pulsations influence intrathecal drug distribution: key factors for interpatient variability
Anesth. Analg.
Computational modeling of the mechanical behavior of the cerebrospinal fluid system
J. Biomech. Eng.
Cited by (73)
Stationary flow driven by non-sinusoidal time-periodic pressure gradients in wavy-walled channels
2023, Applied Mathematical ModellingPatient-specific computational fluid dynamic simulation of cerebrospinal fluid flow in the intracranial space
2022, Brain ResearchCitation Excerpt :As a step towards fulfilling this need, we have developed a three-dimensional, whole brain, transient simulation of CSF flow. The simulation models whole-brain CSF flow using exclusively patient-specific physiological data, without making any boundary condition assumptions for pressure or CSF production/absorption, setting it apart from previous models of CSF flow(Gupta et al., 2010; Khani et al., 2020; Kurtcuoglu et al., 2007; Sweetman and Linninger, 2011; Howden et al., 2011; Tangen et al., 2015). By accounting for all CSF flows in and out of the cranial cavity, we accurately compute CSF velocities and intracranial pressure dynamics without relying on generalized or non-patient-specific assumptions.
Intrathecal delivery and its applications in leptomeningeal disease
2022, Advanced Drug Delivery ReviewsCitation Excerpt :In 2005, Stockman, studied flow in a more anatomically detailed model of the healthy cranial SAS, and found that SAS microstructure leads to pressure drops along the domain [106]. Several groups have since determined that nerve roots and trabeculae substantially alter the velocity field within the CSF, inducing vortices and regions of local mixing that dramatically accelerate drug dispersion [107-110]. CFD modeling has led to other insights: CSF flow was once thought to be fully laminar, but studies using direct numerical simulation have revealed that there are intermittent, localized occurrences of turbulence particularly in disease states [111,112].
Modelling and direct numerical simulation of flow and solute dispersion in the spinal subarachnoid space
2021, Applied Mathematical ModellingCitation Excerpt :With the aim of providing detailed numerical results for validation of the theoretical model, three-dimensional, unsteady direct numerical simulations of the fluid motion and solute dispersion in the canal were carried out. To facilitate the computation, instead of using the intracranial pressure as input, the displacement of the dura membrane was prescribed, that also being the approach of the moving-boundary-motion method first used in the DNS computations of Tangen et al. (2015) [23] (see also Khani et al. 2018 [25]). For consistency, the nonuniform temporal distribution of the dura-membrane displacement was selected to be that found at leading order in the previous theoretical analysis of the fluid-structure interaction problem [28].
Computerized biomechanical simulation of cerebrospinal fluid hydrodynamics: Challenges and opportunities
2021, Computer Methods and Programs in BiomedicineTensile mechanical properties of the cervical, thoracic and lumbar porcine spinal meninges
2021, Journal of the Mechanical Behavior of Biomedical Materials