The Glymphatic System in Cerebrospinal Fluid Dynamics: Clinical Implications, Its Evaluation, and Application to Therapeutics Open Access

The glymphatic system is a macroscopic waste clearance system that eliminates interstitial solutes, including neurotoxic products such as β-amyloid, from the central nervous system (CNS) [1]. The system is currently conceptualized as comprising four segments: (1) periarterial cerebrospinal fluid (CSF) influx, (2) interstitial CSF movement, (3) efflux along the perivenous spaces, and (4) meningeal lymphatic drainage [2] (Fig. 1). Dysfunction in any of these segments may result in the accumulation of neurotoxic waste products, thereby contributing to the pathogenesis of various neurodegenerative diseases [3]. Furthermore, the glymphatic system has been implicated in specific CSF-related conditions, such as idiopathic normal pressure hydrocephalus (iNPH) [4]. Notably, some patients with neurodegenerative diseases develop hydrocephalus, suggesting that impaired CSF dynamics caused by glymphatic dysfunction may underlie certain neurodegenerative diseases presenting hydrocephalus. Clinically, evaluating the glymphatic system is crucial for understanding the pathophysiology of these diseases or potentially predicting disease risk [5]. However, despite the significant demand for such evaluations, studies on the human glymphatic system remain relatively scarce compared to those conducted in rodent models. Although magnetic resonance imaging (MRI) with intrathecal contrast agents is considered the gold standard for evaluating the glymphatic system, alternative imaging modalities, such as diffusion tensor imaging (DTI), have shown promising utility.

The glymphatic system has also emerged as a potential method for enhancing drug delivery. Since systemically injected mid-sized drugs cannot cross the blood-brain barrier (BBB), most therapeutics targeting CNS have been designed for direct injection into the CSF. Given the glymphatic system’s role in CSF dynamics, modulating this system may enhance drug distribution throughout the CNS. For example, the localization of certain drug modalities, such as oligonucleotide therapeutics, within the perivascular space (PVS) after intrathecal injection underscores the role of the glymphatic system in mediating drug distribution across the brain.

This review summarizes the current understanding of the glymphatic system and its dysfunction in relation to specific CSF disorders. Additionally, we highlight advancements in the evaluation of human glymphatic function and explore the potential application of this system for drug delivery.

Before the concept of the glymphatic system was introduced, the dynamics of molecules injected into the CNS had already been investigated. Cserr and colleagues demonstrated the perivascular distribution of the CSF tracer in arteries and veins [6], bulk flow within the PVS [6], and drainage of the tracer into deep cervical lymph nodes [7], following the intracortical injection or subarachnoid space injection. Although they concluded that paravascular influx of CSF tracer was slow and variable in direction [6], these studies likely observed components of the glymphatic system, which was identified later.

In 2012, the concept of a waste clearance pathway in the brain was proposed [1]. Iliff and colleagues demonstrated that CSF tracer or β-amyloid injected into CSF or brain parenchyma was eliminated from the brain through paravenous routes. In 2015, two independent groups discovered lymphatic vessels within the meninges surrounding the brain [8, 9]. The initial segments of the meningeal lymphatic vessels are located near the dural venous sinuses [8, 9], which makes waste products along paravenous space to be eliminated into the meningeal lymphatic vessels. Recent anatomical studies further underscored the significance of meningeal lymphatic vessels located in the basal part of the skull [10] and nasopharyngeal lymphatic plexus [11].

Several factors have been identified to modulate glymphatic function. Arterial pulsation plays a critical role in periarterial CSF influx, the first segment of the glymphatic system. Experimental evidence indicates that fluid flow within the PVS is driven by arterial pulsations derived from the cardiac cycle [12‒17]. Systemic administration of vasoregulators has been shown to affect the glymphatic flow [18], further supporting the role of arterial pulsations in this system. In addition to arterial pulsations derived from the cardiac cycle, arterial dilations during functional hyperemia [19, 20] and slow vasomotion during non-rapid eye movement sleep and intermediate state sleep [21] have been reported to facilitate the glymphatic system. These studies indicated that arterial wall motion associated with functional hyperemia and slow vasomotion may enhance the glymphatic system comparably to cardiac pulsations-mediated arterial wall motion. Another key player in the glymphatic system is aquaporin-4 (AQP4). In the original paper proposing the glymphatic system, AQP4 knockout (KO) mice exhibited a reduced influx of small molecules from the CSF into the brain parenchyma compared to wild-type mice [1]. Subsequent studies corroborated that AQP4, located on the astrocytes, is involved in both perivascular CSF influx and interstitial fluid outflow [22]. AQP4 is a water channel protein predominantly expressed in the astrocytic endfeet membrane at the BBB. It is now widely accepted that AQP4 surrounding the penetrating arteries at least partially drives the CSF influx to the parenchyma, although the precise mechanisms remain to be fully elucidated. Additional factors, including sleep [23], circadian rhythm [24], and anesthesia [25], have also been shown to influence glymphatic function. Although the glymphatic function is thought to be activated during sleep [23] or anesthesia [25], a recent study demonstrated that glymphatic activity is reduced during sleep and under anesthesia [26], warranting further investigation.

Despite increasing evidence supporting the glymphatic hypothesis, there are still some controversies. While the glymphatic system has been well studied in rodent models, evidence of glymphatic function in humans is still limited. Physiological differences between rodents and humans, such as cerebral metabolic rate [27], brain size, and heart rate, may have an effect on glymphatic function in humans as compared to rodents [28]. Therefore, further studies on the human glymphatic system are needed to understand its impact on human CSF dynamics or the pathophysiology of some neurodegenerative diseases. Some studies have raised questions about certain segments of the glymphatic system. Smith and colleagues [29] tested the glymphatic hypothesis using AQP4 KO mice and demonstrated AQP4-independent solute transport, which is inconsistent with the glymphatic hypothesis. However, a follow-up study by five independent groups consistently reported a reduced glymphatic influx in AQP4 KO mice [22]. This study speculated that technical differences, such as anesthetic agents or invasive procedures, may account for these discrepancies. Additionally, Gomolka and colleagues [30] successfully evaluated glymphatic dysfunction in vivo in AQP4 KO mice, further supporting the role of AQP4 in the glymphatic system. It is also important to note that the PVS, where CSF is thought to flow in the current glymphatic models, may be absent after brain fixation [31]. During histological processing, the PVS shrinks or disappears, and CSF tracers are redistributed in the surrounding smooth muscle layer and the basal lamina [16]. This redistribution may lead to potential misinterpretation of CSF tracer localization to be distributed in the smooth muscle layer, which is considered the primary site of waste clearance in the intramural periarterial drainage model [32]. Although it remains inconclusive whether these two pathways (glymphatic system and intramural periarterial drainage) coexist or only one predominates, it is crucial for researchers to recognize the importance of using in vivo imaging techniques to accurately assess the CSF flow dynamics, as employed in the previous study on the glymphatic system [30].

In the traditional CSF flow model, CSF is primarily produced by the choroid plexus within the lateral ventricles. From there, it flows through the foramina of Monro into the third ventricle, continuing through the cerebral aqueduct into the fourth ventricle. CSF then exits the fourth ventricle via three openings: the foramen of Magendie and the two foramina of Luschka, entering the subarachnoid space. Subsequently, it is absorbed into the venous system through arachnoid granulations. This conventional model emphasizes a unidirectional flow of CSF from its production at the choroid plexus to its eventual reabsorption through the arachnoid granulations [33]. In dogs and monkeys, around 30–40% of CSF is produced by the choroid plexus at the lateral and the third ventricles [34‒36], while around 70% of CSF appears to be produced by the choroid plexus at the fourth ventricles in dogs [36]. These direct lines of evidence dictate that the choroid plexus is the primary site of CSF production [37]. However, some groups have addressed the presence of the extrachoroidal CSF. Milhorat and colleagues [38], for instance, reported that bilateral dissection of the choroid plexus led to only a 30% reduction in CSF production, suggesting the presence of extrachoroidal sources of CSF production.

Regarding CSF excretion, Weed and colleagues [39] explored the concept of CSF drainage at the arachnoid granulations. They observed that dye injected into CSF concentrated in the arachnoid granulations above the cerebral hemispheres several hours postinjection, suggesting that arachnoid granulations play a role in CSF efflux. Although the precise role of arachnoid granulations in CSF clearance is not yet fully clarified, some researchers have suggested that arachnoid granulations function as pressure-sensitive valves, which open to flow under elevated intracranial pressure [40]. The relative contributions of this conventional CSF excretion pathway and glymphatic efflux to overall CSF clearance are not yet fully clear. Further research is required to deepen our understanding of CSF dynamics as a whole.

Given its role in regulating CSF dynamics, glymphatic dysfunction is speculated to contribute to CSF disorders. iNPH is a clinical entity characterized by the enlargement of the ventricles and a triad of symptoms: gait disturbance, cognitive impairment, and urinary disturbance [41]. Several studies have explored the relationship between glymphatic dysfunction and the pathophysiology of iNPH (Fig. 2c). For instance, perivascular AQP4 expression is reduced in patients with iNPH compared to those with control diseases [45, 46]. Depolarization of AQP4 can lead to reduced glymphatic flow, potentially resulting in CSF accumulation in iNPH. Ringstad and Eide evaluated glymphatic function in patients with iNPH using MRI with intrathecal contrast agent administration. They reported delayed enhancement in certain brain regions [47] and decreased contrast clearance compared to control subjects, indicating dysfunction in glymphatic influx and efflux in patients with iNPH [47‒49].

Alzheimer’s disease (AD) is the most extensively studied condition in relation to glymphatic system dysfunction. As demonstrated in the seminal work by Iliff et al. [1], numerous studies have shown an association between impaired glymphatic function and the accumulation of β-amyloid [1, 50, 51] or tau [52, 53] in the brain (Fig. 2b). While much of the evidence comes from animal models, several studies in human subjects have also been reported [54, 55]. For instance, Okazawa et al. [54] and Huang et al. [55] utilized the diffusion tensor imaging analysis along the perivascular space (DTI-ALPS) (see Evaluation of the Glymphatic System in Human Subjects for further details) to assess glymphatic activity and found significant correlations with amyloid burden as measured by amyloid positron emission tomography. Although, as discussed later, DTI-ALPS has certain methodological limitations, these findings support the involvement of glymphatic dysfunction in AD pathology in humans.

This growing body of evidence has led to the development of therapeutic strategies aimed at enhancing glymphatic clearance of β-amyloid. Experimental approaches include promoting Aqp4 stop codon readthrough [56], inhibiting AQP4 depolarization via oxytocin [57], stimulating the transient receptor potential vanilloid-4-AQP4 pathway using ultrasound [58], and employing multisensory gamma stimulation to promote both arterial pulsation and AQP4 polarization [59]. Of particular note, multisensory gamma stimulation has already entered clinical trials, highlighting its potential for future clinical use.

Intriguingly, patients with AD may also present with hydrocephalus during disease progression. Autopsy studies of patients clinically diagnosed with iNPH have occasionally revealed AD pathology [60], suggesting the underlying similarities in pathophysiology between AD and iNPH. Moreover, patients with hydrocephalus associated with AD may respond positively to shunt operation [61], a characteristic feature of iNPH. These clinical observations imply that hydrocephalus in AD may result from glymphatic dysfunction, similar to that seen in patients with iNPH.

Parkinson’s disease (PD) is also among the conditions in which glymphatic dysfunction has been reported [62]. In animal model studies, an association has been observed between glymphatic dysfunction and α-synuclein accumulation (Fig. 2b) [63, 64]. In addition, a clinical study has demonstrated glymphatic dysfunction in patients with PD [65]. Similar to AD, hydrocephalus may develop during the disease course in PD [66]. Furthermore, several reports have suggested that, as with iNPH, some patients with PD may respond to shunt surgery [67, 68]. These findings raise the possibility that glymphatic dysfunction contributes not only to α-synuclein accumulation, but also to the development of hydrocephalus. However, no studies to date have directly assessed the relationship between α-synuclein accumulation and glymphatic dysfunction in patients with PD.

Although limited, several clinical studies have investigated glymphatic system function in patients with progressive supranuclear palsy (PSP) [69, 70]. Notably, Jiao et al. [70] employed the DTI-ALPS method to assess glymphatic activity and tau positron emission tomography to evaluate tau accumulation, reporting a significant correlation between glymphatic dysfunction and tau deposition. Additionally, hydrocephalus-like presentations have been reported in PSP. In some cases, patients with pathologically confirmed PSP had exhibited clinical features suggestive of iNPH, leading to antemortem misdiagnosis as iNPH [66, 68, 71‒73]. Our previous studies indicated that PSP patients may develop hydrocephalus more frequently than those with other neurodegenerative diseases [74]. In the prospective study, we also assessed the response to the spinal tap in PSP and found that some patients exhibited clinical improvements beyond placebo effects [75]. However, due to the lack of well-established animal models for PSP [76], a direct relationship between glymphatic dysfunction and tau protein accumulation has not been experimentally demonstrated.

Reactive gliosis, the universal reaction to brain injury, causes glymphatic dysfunction [77]. In a healthy condition, AQP4 expression is highly polarized to the astrocytic endfeet (Fig. 2a); however, this polarized expression is reduced in reactive astrocytes, impairing the glymphatic flow (Fig. 2b, c). Indeed, previous histopathological studies reported that the increasing severity of astrogliosis is associated with altered expression and/or impaired polarization of AQP4 at the astrocytic perivascular endfeet in iNPH [45, 46]. Reactive gliosis is also observed in various neurodegenerative diseases such as AD, PD [78], and PSP [79, 80]. Thus, it is speculated that reactive gliosis may reduce AQP4 polarization at the perivascular astrocytic endfeet, impairing CSF circulation due to glymphatic dysfunction and potentially leading to hydrocephalus in some patients with AD, PD (Fig. 2), and PSP. Future studies are warranted to clarify the underlying glymphatic dysfunction in secondary hydrocephalus patients to neurodegenerative diseases.

Currently, no specific biomarkers for glymphatic dysfunction have been established, and its evaluation in humans relies primarily on imaging techniques. Several studies have assessed the glymphatic function in human patients using various imaging modalities.

MRI has become a pivotal tool in evaluating the glymphatic system due to its noninvasive nature. Although its resolution is insufficient to directly visualize PVS or glymphatic flow, it offers the significant advantage of imaging the entire brain and even the neck region. One approach involves using gadolinium-based contrast agents to trace the movement of CSF through the glymphatic system. A list of dynamic contrast-enhanced MRI for the human glymphatic system published in 2015–present is shown in Table 1. As it is difficult for contrast agents to cross the BBB, intrathecal contrast-enhanced MRI is now considered a gold standard for glymphatic imaging in humans. To date, researchers have used this method to investigate glymphatic function in patients with intracranial hypotension [81], iNPH [47‒49], other various CSF disorders [83‒85], or healthy volunteers [87]. Their research demonstrated that glymphatic flow could be visualized as the contrast agent moved from the subarachnoid space into the brain parenchyma, which is altered by certain disease conditions. Furthermore, the clearance of the CSF tracer through the meningeal lymphatic vessels and/or to the cervical lymph nodes was also documented using intrathecal contrast agents [82, 86]. Although some studies used intravenous gadolinium-based contrast agents, ultrasound-induced BBB damage [88] or BBB dysfunction under the disease condition [89, 90] is required to visualize the glymphatic system.

Safety concerns regarding gadolinium retention within the human CNS after repeated intravenous administrations have been discussed [91‒93]. A systematic review concluded that adverse events occur in a dose-dependent manner, with serious neurotoxic complications associated with intrathecal gadolinium-based contrast agent doses of 1 mmol or higher [94]. Interestingly, the severity of these neurotoxic events was not consistently correlated with the administered doses. It is also important to note that most cases included in the systematic review involved gadopentetate dimeglumine, a linear gadolinium-based agent. In general, linear gadolinium-based contrast agents, such as gadopentetate dimeglumine, are associated with greater tissue deposition and longer retention times compared to macrocyclic gadolinium-based agents, such as gadobutrol [95]. Although no direct evidence currently demonstrates harmful effects of gadolinium deposition in the brain, clinicians might consider that some linear agents exhibit a higher propensity for deposition than some macrocyclic agents [91, 94, 96, 97]. Indeed, the use of linear agents has been restricted based on the potential risk of brain deposition in Europe [98]. Future studies are needed to confirm the safety of gadolinium-based contrast agents in the CNS and the differences between the two types of agents from the safety perspectives.

An alternative approach other than gadolinium-based contrast agents is using ¹⁷O-labeled water. ¹⁷O is the only stable isotope of oxygen that produces signal changes on MRI [99]. Unlike contrast agents, this technique directly labels water, which allows us to precisely evaluate the water dynamics in the brain. Kudo and colleagues [100‒102] explored the feasibility of indirect proton MRI and kinetic analysis of ¹⁷O-labeled water in animals and humans. This method involves administering ¹⁷O-labeled water intravenously and then performing dynamic steady-state MRI scans to track the distribution and kinetics of ¹⁷O-labeled water in the brain. Their findings indicated that ¹⁷O-labeled water could effectively trace CSF dynamics and provided insights into its permeability to the brain [101, 102]. However, this unique technique using ¹⁷O-labeled water has several limitations. The spatial resolution is not high enough to directly evaluate the arterial signals [101] or glymphatic flow. Furthermore, a substantial quantity of ¹⁷O-labeled water is required for intravenous administration, which may result in significant cost-related challenges [99].

DTI is an advanced MRI modality that visualizes the diffusion of water molecules. DTI-ALPS is a noninvasive method using diffusion MRI to assess the human glymphatic system [103]. DTI-ALPS evaluates the diffusivity of water along the PVS at the level with the upper part of the lateral ventricles where the PVS runs perpendicular to major white matter tracts [103]. This method calculates the ALPS index, representing the ratio of the water diffusivity in PVS along the x-axis (perpendicular to projection and association fibers) to the water diffusivity along the y- and z-axis (parallel to these fibers). This ratio reflects relative fluid movement through the PVS, thereby assessing the glymphatic activity to some extent. They found significant correlations between the ALPS index and the mini-mental state examination score in patients with AD [103], suggesting this method’s utility in assessing the glymphatic activity in CNS diseases. To date, many researchers have employed the DTI-ALPS to evaluate the glymphatic system in various CNS diseases, such as AD [104, 105], PD [106‒108], iNPH [109], traumatic brain injury [110, 111], Moyamoya disease [112], and PSP [69, 70].

However, there remains controversy in the DTI-ALPS method [113‒116]. Only one paper has validated the relationship between the ALPS index and glymphatic activity by comparing it with gold-standard contrast-based glymphatic imaging [117]. Moreover, DTI-ALPS estimates water diffusion in relatively deep brain regions, while contrast-based glymphatic imaging mainly evaluates glymphatic activity near the cortex. Thus, the evaluation of water diffusion in deep brain regions may reflect only a part of the whole glymphatic activity [116, 118]. Taken together, the decreased ALPS index should not be simply interpreted as impaired glymphatic dysfunction [119].

Given its superior spatial resolution, 7-Tesla (7T)-MRI is increasingly utilized for noninvasive assessment of the human brain [120]. Notably, it has enabled the visualization of fine anatomical structures that are difficult to detect with lower field strengths [120, 121]. For instance, Patel et al. [121] employed contrast-enhanced 7T-MRI to visualize meningeal lymphatic vessels and identified their locations along the superior sagittal sinus and cortical veins. Thus, 7T-MRI holds promise for directly visualizing components of the glymphatic system in humans. However, even at this high resolution, penetrating arteries extending from the cortex, whose periarterial spaces are considered part of the glymphatic CSF pathway, remain difficult to visualize [122]. As 7T-MRI is still an emerging technology, further studies and evidence accumulation are warranted.

The discovery of the glymphatic system has also brought novel therapeutic insights to CNS diseases. This review focuses on the impacts of modulating the glymphatic system/CSF dynamics on drug delivery.

Similar to small molecules injected into CSF [1], many drugs targeting CNS diseases are distributed through the glymphatic pathway. Antisense oligonucleotides (ASOs) are colocalized with AQP4 at the penetrating arteries in the mouse cortex after intracisterna magna injection [123]. Besides, intracisterna magna injection of ASOs to AQP4 KO mice showed lesser ASO distribution and efficacy than wild-type mice [123]. siRNAs, another oligonucleotide drug, are also distributed along the PVS [124], suggesting the role of the glymphatic system in siRNA distribution. Moreover, the distribution of adeno-associated viruses throughout the CNS may also be affected by the glymphatic system. Murlidharan and colleagues [125] reported that intraventricular injection of adeno-associated viruses to AQP4 KO mice showed better distribution and efficacy than wild-type mice. Based on these reports, it is possible to enhance the delivery of therapeutic agents to the CNS by modulating the glymphatic system.

Several studies have demonstrated methods for modulating the glymphatic system to enhance drug delivery (Table 2). Ultrasound is one of the methods to manipulate the glymphatic system [58, 132‒134]. To date, several papers have reported that ultrasound waves with [132, 134] or without microbubbles [58, 133] may yield interstitial convection flow in the brain, which enhances the glymphatic system. Aryal and colleagues [126] applied ultrasound for drug delivery and demonstrated the utility of this method for enhancing intrathecal drug distribution throughout the brain. The ultrasonic treatment enhanced the brain distribution of even a large molecule, a fluorescence-labeled monoclonal antibody (∼155 kDa). The systemic administration of hypertonic saline is one of the other methods to modulate the glymphatic function [127‒129, 135]. Increased plasma osmolality due to hypertonic saline draws water from the brain and spinal cord to the systemic circulation. The loss of water in the brain and spinal cord facilitates the perivascular influx of CSF along penetrating arteries, following along the pressure gradient, which enhances the glymphatic system. The brain distribution of small gold nanoparticles and antibodies injected into CSF can be improved by systemic hypertonic saline [127, 129] or intrathecal mannitol infusion [130]. Some anesthesia drugs may also modulate the glymphatic system. Since α₂-adrenergic agonists, dexmedetomidine or xylazine, have been reported to enhance the glymphatic system relative to isoflurane and awake state [23, 25, 136], these anesthesia drugs can improve drug delivery [131].

The discovery of the glymphatic system has brought us new insights into the pathophysiology of CSF disorders. Although human evidence of this system is still limited, some imaging studies have revealed close relationships between glymphatic dysfunction and CSF disorders. Furthermore, the glymphatic system’s potential for clearing pathogenic protein aggregations or enhancing drug delivery to the brain offers new therapeutic opportunities for CNS disorders. Future research should continue to explore and optimize these applications, potentially transforming the landscape of neurodegenerative disease treatment.

Masahiro Ohara has no conflicts of interest to declare. Takaaki Hattori has received speaker’s honoraria from Daiichi Sankyo Co., Limited; Sumitomo Dainippon Pharma Co., Ltd.; Integra Japan Co., Ltd; and Kyowa Kirin Co., Ltd., and was a member of the journal’s Editorial Board at the time of submission.

Any sponsor or funder did not support this study.

M.O. drafted the original manuscript and created the figures. T.H. revised and edited the manuscript.