The space between the arachnoid and pia meningeal layers encasing the brain is a landscape of connective tissue, blood vessels, and cerebrospinal fluid. Scientists debate how that fluid moves within the space and through brain tissue. Now, an MRI study by Per Kristian Eide and Geir Ringstad of the Oslo University Hospital in Norway supports the idea that people have a glymphatic system. In the March 5 Nature Communications, they reported serial scans showing CSF flowing into the brain by hugging cerebral arteries in the subarachnoid space. Within minutes, the fluid seeped into adjacent brain tissue. This flow was slowed in people with hydrocephalus, a disorder that causes a buildup of fluid in the brain.

  • In people, CSF flows into the brain around arteries.
  • Within minutes, it percolates into the parenchyma.
  • Hydrocephalus slows this movement.

“This paper shows that the subarachnoid space is divided into different compartments—the perivascular space and the outer subarachnoid space,” said Maiken Nedergaard of the University of Rochester Medical Center in New York. “This is important for glymphatic clearance, enabling clean CSF to flush in, and dirty fluid to exit, separately.”

The glymphatic clearance, a lymph-like process that relies on astrocytes to wash CSF through the brain, has been documented in mice. The fluid flows into brain tissue through perivascular spaces along arteries and drains out along veins (Aug 2012 news). Supporting this, a recent mouse imaging study showed CSF exiting the brain along small bridging veins. It slipped out the arachnoid membrane through openings called “cuffs” around these vessels into the dura mater and on into lymph vessels (Feb 2024 news).

Previously, Eide and Ringstad had focused on CSF movement within the brain, tracking an MRI tracer as it flowed next to arteries without penetrating them (Oct 2022 news). To map how CSF entered the brain, Eide analyzed consecutive MRI scans from 75 adults intrathecally injected with the tracer gadobutrol. Of these participants, 14 were healthy, 22 had idiopathic normal-pressure hydrocephalus (iNPH), and the rest had other CSF disorders, such as arachnoid or pineal cysts, or intracranial hypo- or hypertension. Participants were 50, on average, and two-thirds were women.

CSF Doughnut. In the subarachnoid space (SAS) of the cerebral cortex (CC), gadobutrol (white, top left; blue, top right) gathers in the perivascular subarachnoid space (PVSAS) around an artery (A). Three-dimensional models (bottom) of the MRI scans depict the pia (P), perivascular membrane (PVM), and the arachnoid trabeculae (AT)—strands of connective tissue connecting the arachnoid and pial membranes. [Courtesy of Eide et al., Nature Communications, 2024.]

Over an average of 14 minutes, the tracer traveled from the injection site in the lumbar spine to the basal cisterns, i.e., CSF reservoirs, at the base of the skull. Then, it concentrated around the anterior, middle, and posterior cerebral arteries, forming perivascular rings (image at right).

After the tracer surrounded arteries, it followed them into the parenchyma. Within minutes, the tracer diffused into the cerebrum into which the arteries feed, permeating the frontal cortex, and temporal cortex. Two hours after injection, the tracer had seeped deep within the brain (image below). That the tracer ended up in cortical tissue suggested to Eide that this CSF inflow feeds into the glymphatic system.

Because there was a delay before the tracer percolated from the perivascular space into the parenchyma, the authors believe that a semi-permeable membrane encircles the subarachnoid space around arteries to create CSF channels. However, the low resolution of MRI prevented them from seeing this membrane. Researchers led by the late Roy Weller had proposed such a membrane might regulate retrograde flow of CSF from the brain along arteries, but Eide and Ringstad believe that the flow must be antegrade (Zhang et al., 1990).

Creeping Up. An MRI tracer enters the basal cisterns (left panel, white area at bottom), then flows through perivascular subarachnoid space (PVSAS) next to the anterior cerebral artery (A2) into the brain, seeping into the parenchyma. Asterisks indicate the tracer’s appearance at different spots along the artery. [Courtesy of Eide et al., Nature Communications, 2024.]

Might CSF flow change in people with brain disorders? Indeed, the fluid trekked sluggishly in people with normal-pressure hydrocephalus, despite their having enlarged perivascular subarachnoid spaces. The tracer took 15 minutes longer to appear at the anterior cerebral artery than it did in controls, and 2.5-fold less trickled into the cortex two hours after injection. The authors chalk this up to high intracranial pulse pressure from enlarged ventricles. They think this restricts artery pulses, which slow perivascular CSF movement, as the pulses gently push the fluid along.

Notably, the scientists rarely saw the tracer around veins. Though veins are slightly harder to see on MRI than arteries, Eide does not think this is simply a methodological issue. “If there was perivenous tracer enrichment, I expect we would see it,” he told Alzforum. This made sense to Nedergaard, who also sees no CSF outflow in mouse cortical veins, instead spotting it in deep, central veins near the neck. Because veins do not pulsate as strongly as arteries, she thinks the cortical perivenous spaces might not have enough pressure to push CSF out.—Chelsea Weidman Burke


  1. This is an exciting study—it provides the first view into tracer movement through the human subarachnoid space over just a few hours. They find that tracers move in a specific pathway along the arteries before entering brain tissue. This information is important to understand the pathways of waste clearance in the human brain, which have been very challenging to image due to the tiny, intricate structures involved.

  2. Divide and Conquer: Compartmentalization of the Brain’s Subarachnoid Space Facilitates Rapid Periarterial CSF Flow

    Over the past decade, there has been significant interest in comprehending the underlying anatomy and physiological mechanisms that regulate the exchange of cerebrospinal fluid (CSF) and interstitial fluid (ISF) within the brain. Throughout this period, the initial characterization of the glymphatic pathway by Nedergaard and colleagues (Iliff et al., 2012), demonstrating the swift movement of CSF into the brain parenchyma through periarterial spaces and the clearance of ISF within perivenous spaces, has been repeatedly replicated by many groups in multiple model systems. While most of these studies have been conducted in non-human subjects, there is a growing body of research, including contributions from Eide and Ringstad's group, validating the existence and function of the glymphatic pathway in humans using non-invasive MRI-based techniques (Eide and Ringstad, 2015; Ringstad et al., 2017; Eide et al., 2018; Ringstad and Eide, 2022). 

    In this article by Eide and Ringstad, the authors delve deeper into characterizing the anatomical and functional aspects of leptomeningeal perivascular CSF flow dynamics in humans under both healthy and diseased conditions. They utilize intrathecal gadobutrol-enhanced T1-weighted MRI and report the compartmentalization of the human subarachnoid space into what they propose is an inner perivascular subarachnoid space (PVSAS) and an outer generalized subarachnoid space (SAS). The PVSAS predominantly surrounds major arteries from the circle of Willis, such as the anterior cerebral artery (ACA), middle cerebral artery (MCA), and posterior cerebral artery (PCA). The authors observe direct CSF solute propagation between the basal cisterns and the PVSAS, and confirm antegrade CSF flow within the PVSAS of large surface arteries before reaching the adjacent cerebral cortex. Functionally, they note slowed CSF tracer appearance in periarterial subarachnoid spaces with increased pulsatile intracranial pressure, suggesting potential impairments in diseases like communicating hydrocephalus. Additionally, they find an enlarged PVSAS area in idiopathic normal-pressure hydrocephalus (iNPH), associated with altered CSF and solute flow dynamics and tracer accumulation in specific cortical regions.

    The study by Eide and Ringstad deserves commendation for its impact, relevance, and methodological rigor. Nevertheless, it raises crucial questions, particularly regarding the membrane that segregates the inner perivascular space from the broader subarachnoid space. While the authors propose this membrane is part of the arachnoid, they provide limited data supporting an arachnoid rather than pial identity. Counter to this conception of a subarachnoid perivascular space, previous anatomical studies suggest that surface arteries and veins run in a sub-pial plane, with Virchow-Robin spaces formed by pial membrane invagination as surface arteries penetrate the brain parenchyma (Zhang et al., 1990). Further, recent work from Nedergaard's group identifies a potential 4th meningeal membrane, termed the subarachnoid lymphatic-like membrane (SLYM), distinct from the arachnoid, based on its expression of lymphatic endothelial cell markers (LYVE-1, PROX-1, and PDPN) (Plá et al., 2023). Consequently, it is also possible that this SLYM layer demarcates perivascular spaces from the subarachnoid space. Future investigations focusing on labeling for arachnoid barrier cells, pial cells, and lymphatic markers in postmortem human samples may clarify this membrane's identity.

    Moreover, Eide and Ringstad's study reveals that tracer initially appears in the PVSAS of major arteries near the circle of Willis, suggesting CSF entry at proximal locations within the basal cisterns. The exact anatomical site of this entry, and the cellular and molecular components forming it, remain unknown, warranting further exploration. Further, this perivascular membrane appears to function akin to myelin on an axon in promoting rapid axial CSF and solute movement down arteries, while restricting radial movement into the broader SAS. The mechanisms underlying this barrier function, however, remain elusive. If this membrane indeed consists of arachnoid barrier cells, it may contain tight junction proteins that prevent perivascular contents from entering the surrounding SAS. Over time (three hours), there appears to be signal equilibration between the PVSAS and the SAS, leading the authors to suggest that the membrane is semipermeable. Supporting this, they demonstrate rapid signal enrichment in the adjacent cortex after tracer appearance in the PVSAS. This suggests similarities to fenestrated pia in Virchow-Robin spaces (Zhang et al., 1990) or SLYM, which allows fluid and solute movement up to 3 kDa in size (Plá et al., 2023). Future research should focus on understanding the factors contributing to this membrane's semipermeability, such as molecular weight or charge.

    While most of the tracer is concentrated around arteries, there is also signal, though less pronounced, around surface veins (see supplemental figure 4). This hints at possible communication between the PVSAS of arteries and veins or the direct entry of tracer into the perivenous subarachnoid space from the basal cisterns. The latter scenario seems less probable, given prior research indicating that CSF and tracer predominantly flow toward the brain through periarterial spaces rather than perivenous spaces (Iliff et al., 2012). Furthermore, it is unlikely that tracer is entering the perivenous space directly from the parenchyma (the clearance side of the glymphatic pathway) at such an early stage (less than one hour). Therefore, there is likely direct solute communication from periarterial spaces to adjacent perivenous spaces, similar to what is observed in the adjacent cerebral cortex. This raises similar questions to those previously mentioned regarding factors limiting the movement of CSF and solute across this semipermeable membrane. Is the limited tracer observed in the perivenous space due to characteristics of the tracer molecule itself, such as its size? Would a larger molecule, exceeding the size of gadobutrol (690 Da), face greater restriction entering the perivenous space, or would its size hinder its movement into the brain parenchyma, causing it to accumulate in the perivenous space? These questions necessitate further thorough investigation, as their answers may provide insights into why proteinopathies such as cerebral amyloid angiopathy predominantly affect arteries rather than veins. Additionally, it would be intriguing to expand upon the remarkable temporal resolution showcased in this study to a range of 3-6 hours. Previous research conducted by the authors' group has identified enhancements in tracer activity around bridging veins and within the parasagittal dura (Ringstad and Eide, 2022). Recent findings from our own group indicate that arachnoid barrier discontinuities, referred to as arachnoid cuff exit (ACE) points, coincide with the entry of these bridging veins into the dura. This phenomenon allows CSF to permeate into the dura, playing a crucial role in CSF efflux (Smyth et al., 2024). 

    In conclusion, Eide and Ringstad's work presents the first human in vivo MRI evidence of periarterial space segregation, facilitating rapid CSF and solute movement from basal cisterns to adjacent cerebral cortex. Yet, key questions persist regarding the identity of the compartmentalizing membrane, the etiology of this membrane’s semipermeability, and the site of CSF entry into perivascular spaces. Future studies probing these aspects may illuminate fundamental mechanisms impacting brain fluid dynamics in health and disease.


    . A paravascular pathway facilitates CSF flow through the brain parenchyma and the clearance of interstitial solutes, including amyloid β. Sci Transl Med. 2012 Aug 15;4(147):147ra111. PubMed.

    . MRI with intrathecal MRI gadolinium contrast medium administration: a possible method to assess glymphatic function in human brain. Acta Radiol Open. 2015 Nov;4(11):2058460115609635. Epub 2015 Nov 17 PubMed.

    . Glymphatic MRI in idiopathic normal pressure hydrocephalus. Brain. 2017 Oct 1;140(10):2691-2705. PubMed.

    . Magnetic resonance imaging provides evidence of glymphatic drainage from human brain to cervical lymph nodes. Sci Rep. 2018 May 8;8(1):7194. PubMed.

    . Molecular trans-dural efflux to skull bone marrow in humans with CSF disorders. Brain. 2022 May 24;145(4):1464-1472. PubMed.

    . Interrelationships of the pia mater and the perivascular (Virchow-Robin) spaces in the human cerebrum. J Anat. 1990 Jun;170:111-23. PubMed.

    . Structural characterization of SLYM-a 4th meningeal membrane. Fluids Barriers CNS. 2023 Dec 14;20(1):93. PubMed.

    . Identification of direct connections between the dura and the brain. Nature. 2024 Mar;627(8002):165-173. Epub 2024 Feb 7 PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Brain Drain—“Glymphatic” Pathway Clears Aβ, Requires Water Channel
  2. Meningeal Cuffs Around Veins Form Exit and Entry Ramps to the Brain
  3. In Hydrocephalus, Slow Drainage May Cause Dementia

Paper Citations

  1. . Interrelationships of the pia mater and the perivascular (Virchow-Robin) spaces in the human cerebrum. J Anat. 1990 Jun;170:111-23. PubMed.

Further Reading

No Available Further Reading

Primary Papers

  1. . Functional analysis of the human perivascular subarachnoid space. Nat Commun. 2024 Mar 5;15(1):2001. PubMed.