Cerebrospinal Fluid Dynamics and Intracranial Volume Regulation

Cerebrospinal fluid is a regulated extracellular fluid within a connected mechanical system. Its volume reflects secretion, exchange, outflow, and any artificial drainage, while its movement includes large oscillations that need not represent net transport. Understanding these distinctions connects epithelial physiology to pressure regulation, hydrocephalus, and the interpretation of a draining catheter.

61.1 Fluid Compartments Communicate Without Becoming Identical

Cerebrospinal fluid, or CSF, occupies the ventricles and subarachnoid spaces surrounding the brain and spinal cord. Brain interstitial fluid occupies extracellular spaces within tissue. These fluids communicate, but their anatomical locations, access to blood, local composition, and transport constraints differ.

CSF provides mechanical support and a medium for transporting solutes and signaling molecules. Its composition matters as much as its volume. Neuronal function depends on controlled extracellular potassium, hydrogen ion activity, and other solutes, so the system cannot be understood simply as water entering and leaving a container.

The relevant boundaries include the cerebral vascular endothelium, choroid plexus epithelium, ventricular lining, pial and perivascular interfaces, and arachnoid barrier. These structures have different permeability and transport properties. A molecule crossing one interface has not necessarily gained unrestricted access to every other compartment.

The choroid plexus is a major site of regulated CSF secretion, but water and solutes also exchange across other interfaces. Distinguishing an exchange flux from net fluid formation is essential. Large movements in both directions can coexist with a much smaller net addition to CSF volume.

Likewise, the ventricular and subarachnoid fluid spaces form a connected system rather than a uniform mixing vessel. Pathway geometry, regional restrictions, posture, and pulsatile displacement influence how quickly pressure and solute changes are transmitted. A sample from a ventricular catheter need not reproduce the instantaneous composition of lumbar CSF or local brain interstitial fluid.

Volume regulation also occurs within the constraint developed in Chapter 58. CSF can redistribute into connected spinal spaces or leave through outflow pathways, allowing changing blood or tissue volume to be accommodated. Redistribution within the craniospinal system changes regional volume without necessarily changing the total amount of CSF.

Three questions should therefore remain separate: how much fluid is present, where it is moving, and which solutes are being transported. The same observation rarely answers all three.

61.2 The Blood–CSF Barrier Is an Epithelial Transport Interface

The principal blood–brain barrier resides at the endothelium of most cerebral microvessels. In the choroid plexus, the capillary endothelium is fenestrated, allowing relatively ready access from blood to the stromal compartment. The principal blood–CSF barrier instead lies at the epithelial cells surrounding that stroma.

Tight junctions between choroid epithelial cells restrict paracellular passage. Water and selected solutes can still cross through regulated transcellular routes. A barrier is therefore compatible with active secretion: restricting uncontrolled movement is what permits the epithelium to establish selective transport and a regulated ventricular environment.

Brightman and Reese's ultrastructural tracer experiments distinguished the junctions of cerebral vascular endothelium and choroid epithelium from the more permissive intercellular relationships elsewhere. Their findings supplied a structural basis for separating the blood–brain and blood–CSF barriers. They did not establish that all interfaces behave as impermeable walls. 1

Much of the ordinary ventricular ependymal lining lacks the continuous restrictive junctional arrangement of choroid epithelium. Exchange between ventricular fluid and adjacent extracellular spaces is consequently possible, although specialized regions and developmental states require qualification. An epithelial barrier at the plexus should not be mentally extended over the entire ventricular surface.

The barrier also separates hydrostatic access from final fluid composition. Plasma reaching choroid stroma has not simply become CSF. Epithelial ion transport, selective solute entry and removal, electrical gradients, and water permeability intervene before fluid enters the ventricle.

Under usual conditions, CSF has much less protein than plasma, while concentrations of individual electrolytes reflect active homeostasis rather than a uniform dilution of blood. Disease can alter these relationships through inflammation, hemorrhage, or disruption of transport. A change in protein concentration can also reflect altered clearance or mixing, so it does not uniquely identify a barrier lesion.

The useful physiological model is a polarized, metabolically active interface. Its permeability, transport direction, and energy supply jointly determine what crosses it.

61.3 Active Ion Transport Supports Choroid Plexus Secretion

Choroid epithelial cells have a ventricular-facing apical membrane and a blood-facing basolateral membrane. Their polarized transport machinery coordinates ion entry, recycling, and exit across these surfaces. Several key proteins have an unusual distribution compared with many peripheral secretory epithelia.

The sodium–potassium ATPase is prominent on the apical membrane. Each ATP-dependent transport cycle moves three sodium ions out of the cell toward CSF and two potassium ions into the cell. The resulting intracellular ionic conditions support secondary transport and help regulate ventricular potassium as well as sodium movement.

Blood-facing bicarbonate-coupled sodium entry and chloride–bicarbonate exchange help replenish intracellular solutes. Ventricular-facing sodium–bicarbonate transport, chloride pathways, and potassium recycling participate in the integrated transepithelial flux. Carbonic anhydrase accelerates the reversible interconversion of carbon dioxide and water with hydrogen and bicarbonate, supporting substrate supply and intracellular acid–base conditions. These channels, exchangers, and cotransporters work as a network rather than independent pumps each generating a separate fluid stream. Their roles are supported by experimental transport and expression studies, with species and preparation affecting quantitative interpretation. 2

Water movement accompanies this coordinated solute transport. Local gradients near transporting membranes can matter even when bulk CSF and plasma osmolalities are similar. Water-channel permeability, paracellular properties, and proposed transporter-coupled water movement must therefore be distinguished from the assertion that a large sustained bulk osmotic difference is required.

Pollay and colleagues combined ventricular perfusion with measurements of choroid ATPase activity in rabbits and dogs. Inhibition of sodium–potassium ATPase markedly reduced measured intraventricular fluid formation in that preparation. This supports energy-dependent transport, without specifying the fraction of secretion attributable to one mechanism in every species or physiological state. 3

Steffensen and colleagues subsequently used mouse in vivo measurements and isolated plexus experiments to support a role for the apical sodium–potassium–chloride cotransporter, NKCC1, including transporter-associated water movement. Their quantitative contribution estimate belongs to those experimental conditions and is not a measured adult-human production fraction. 4

Adequate perfusion and metabolism sustain these processes, but secretion is not a fixed fraction of cerebral blood flow. Changes in substrate delivery, transporter activity, inflammation, and local chemical conditions can alter the relationship.

61.4 Transport Direction Depends on the Ionic State and Development

NKCC1 moves sodium, potassium, and chloride together, with one sodium ion, one potassium ion, and two chloride ions per transport cycle. Because the charges balance, its cotransport is electrically neutral. Direction depends on the combined chemical gradients of its transported ions, rather than membrane voltage alone.

An apical location therefore does not establish whether the transporter adds solute to CSF or retrieves it. Changing extracellular potassium, intracellular chloride, or pump activity can change the net driving conditions. Transporter abundance and phosphorylation influence activity, while the ionic gradients determine the direction favored by the coupled process.

Gregoriades and colleagues studied mouse choroid epithelial cells using genetic deletion, pharmacological inhibition, imaging, and intracellular ion measurements. Their findings supported net inward NKCC1 activity under the basal conditions studied, maintaining cellular chloride and water volume. This illustrates why cell preparation and measured gradients matter when interpreting an apparent secretory or absorptive role. 5

Development adds another distinction. Xu and colleagues found that early postnatal mice had a transient CSF potassium environment associated with choroid potassium and water clearance. Increasing choroid NKCC1 expression enhanced clearance and altered fluid-volume behavior in that developmental setting. This is not evidence that the same pathway normally removes a specified fraction of CSF in an adult intensive care patient. 6

In later adult-mouse work, Jensen and colleagues combined choroid-specific transporter manipulations with in vivo fluid measurements and ex vivo potassium handling. NKCC1 knockdown reduced secretion, whereas altered potassium conditions favored inward transport and potassium clearance without necessarily producing a corresponding reduction in total secretion. AQP1 disruption did not reduce secretion in their experimental system, emphasizing that a water channel's presence does not establish its quantitative necessity under every condition. 7

These findings should be interpreted at the level actually measured: ionic flux, cell volume, net fluid secretion, or whole-system accumulation. They are not interchangeable endpoints. The evidence supports a regulated epithelial system with context-dependent behavior, not a universal one-direction diagram or a proven clinical drug strategy.

61.5 Net Formation Is a Mass-Balance Quantity

The amount of CSF changes according to the balance of inputs and outputs. A useful accounting equation is:

VCSF is total CSF volume in mL, and t is time in minutes. QChP is net choroid secretion, Qexchange is net entry from other communicating fluid compartments, Qout is native outflow, and Qdrain is artificial drainage; each rate is expressed in mL/min. Qexchange can be negative if net exchange is directed out of CSF. The compartment boundaries must be specified so the same flux is not counted twice.

For a constructed example, assume net choroid secretion of 0.30 mL/min, no net exchange term, native outflow of 0.25 mL/min, and no artificial drainage. The instantaneous balance is 0.30 − 0.25 = 0.05 mL/min. If those rates were artificially held constant for 120 minutes, accumulation would be 0.05 × 120 = 6 mL.

This is an accounting demonstration, not a prediction of how much volume will accumulate safely. Rising pressure can increase native outflow, alter exchange, and change the other conditions. The pressure response also depends on the nonlinear reserve described in Chapter 58. A persistent imbalance cannot be projected indefinitely with all feedback mechanisms held fixed.

Measuring formation is more difficult than writing its balance. Ventricular perfusion and tracer-dilution methods estimate rates from changes in marker concentration under assumptions about mixing, tracer retention, alternative exits, and exchange. A tracer leaving the sampled space can be mistaken for dilution by newly secreted fluid if its loss is not accounted for.

Likewise, flow through the aqueduct is not automatically total secretion: fluid may move bidirectionally, originate at multiple sites, and leave through routes outside that measurement plane. Fluid collected from a drain is one output stream, not a direct secretion meter.

Every reported rate therefore requires its measurement method and physiological conditions. A plausible numerical value does not remove those assumptions.

61.6 The Craniospinal Pathways Form a Connected Network

In the familiar ventricular route, lateral ventricles communicate with the third ventricle through the interventricular foramina, the third with the fourth through the aqueduct, and the fourth with subarachnoid spaces through its outlets. Cranial cisterns communicate with spinal subarachnoid spaces and with fluid around the cerebral surface.

This anatomy establishes possible routes, but it does not mean each fluid parcel travels once around a fixed circuit. Mixing, oscillation, local exchange, and parallel exits complicate the net transport pattern. A tracer's sequence of appearance depends on those processes as well as on the access route used to introduce it.

Connected spinal spaces also provide mechanical accommodation. During a transient increase in intracranial vascular volume, CSF can be displaced caudally, with subsequent return as the vascular volume change reverses. Cranial volume changes while total craniospinal CSF volume may remain nearly unchanged.

The distribution of that accommodation depends on spinal and cranial compliance, venous conditions, and pathway resistance. Posture changes hydrostatic gradients, while respiration changes venous and thoracic pressures. These modifiers can change regional displacement without changing choroid secretion over the same short interval.

A restriction separates the network functionally. Aqueductal narrowing can limit exchange between upstream ventricles and downstream spaces. Outlet or cisternal obstruction produces a different distribution. A pathway that transmits a slow pressure change may still impede rapid pulsatile movement, so communication is not simply an all-or-none property.

The same principle limits inference from a pressure measured at one site. A ventricular pressure, lumbar pressure, and local tissue pressure need not agree when pathways are obstructed or compartments are distorted. Establishing that fluid spaces are anatomically connected does not prove that they are at the same pressure at every moment.

Anatomical imaging and dynamic measurements thus address complementary questions: which routes exist, how restrictive they are, and what is being transmitted through them. A description of CSF dynamics should retain all three.

61.7 Pulsatile Displacement Is Not Net Bulk Flow

Cardiac and respiratory changes repeatedly displace CSF. Arterial expansion, venous volume shifts, and pressure changes across the craniospinal axis generate oscillatory movement. Much of a fluid volume moving one way during part of a cycle can return during another part.

Net flow is the signed volume crossing a specified plane over a complete interval, divided by that interval's duration. Oscillatory stroke volume describes displacement during part of a cycle. Adding forward displacement alone measures something different from subtracting the returning volume.

Suppose a hypothetical measurement plane shows 0.08 mL passing caudally and 0.07 mL passing cranially per cardiac cycle, at 60 cycles/min. Caudal passage sums to 4.8 mL/min and cranial passage to 4.2 mL/min. The signed net is only (0.08 − 0.07) × 60 = 0.60 mL/min caudally.

The example demonstrates subtraction of bidirectional displacement; it is not a normal CSF-flow value or a formation estimate. When two relatively large measured quantities nearly cancel, small offsets in velocity or segmentation can strongly influence the residual. Respiratory sampling and averaging duration can change it further.

Dreha-Kulaczewski and colleagues used real-time MRI in healthy human subjects and found a strong inspiratory influence on CSF motion, particularly during forced breathing. Their result establishes the importance of respiratory conditions in the studied measurements, rather than proving that cardiac pulsation is negligible in every location or physiological state. 8

Oscillations can still contribute to transport without producing a large net volume stream. Repeated displacement can enhance mixing and dispersion, bringing solutes into regions from which diffusion or other pathways redistribute them. Thus, little net movement across one plane does not mean no molecular exchange.

Conversely, vigorous visible motion does not establish adequate outflow or clearance. A closed or restricted system can oscillate while retaining fluid or solute. MRI velocity, net transport, secretion, and clearance therefore require distinct interpretations.

61.8 Outflow Uses Multiple Pathways and Driving Conditions

The classical outflow description emphasizes arachnoid villi and granulations communicating with venous structures. Other routes include cranial and spinal perineural pathways and connections to meningeal and extracranial lymphatic systems. Their relative contribution can depend on species, age, pressure, anatomical access, and the material being followed.

Fluid leaves through a pathway when the relevant driving forces favor movement and the route is sufficiently conductive. For a pressure-driven route, the difference between CSF pressure and downstream pressure matters more than either pressure alone. A change in venous or lymphatic downstream conditions can therefore alter outflow even if CSF pressure initially remains unchanged.

It is useful to imagine parallel routes with different conductances and opening behavior. Blocking one can redirect flow through another, provided a sufficient gradient develops. A single aggregate outflow resistance can summarize that behavior over a limited range but does not identify the individual anatomical route responsible.

Ringstad and Eide followed intrathecal tracer with MRI in 18 people investigated for suspected CSF disorders. Tracer reached parasagittal dura near cortical venous entry regions, supporting trans-arachnoid molecular passage and a possible link to dural lymphatic drainage. The imaging did not quantify what fraction of total bulk CSF outflow used that pathway. 9

Albayram and colleagues combined contrast-enhanced MRI in five healthy participants with ex vivo human tissue mapping. They identified organized lymphatic-like architecture around the middle meningeal artery and delayed enhancement patterns consistent with a distinct compartment. Intravenous contrast kinetics and anatomical markers support further investigation; they do not independently establish a complete volumetric CSF clearance budget. 10

The distinction between anatomical evidence, tracer passage, and quantified bulk outflow is fundamental. An observed molecule may move by diffusion, local exchange, or transport accompanying fluid. Its appearance in a distant structure does not determine how much water followed it or what proportion of physiological clearance that route provides.

61.9 Resistance Sets Flow Requirements; Compliance Sets Storage Responses

For a simplified steady-state system with one effective pressure-driven outflow route, a familiar relationship is:

PCSF and effective downstream pressure, Pout, are in mmHg. Net formation, Qformation, is in mL/min. Outflow resistance, Rout, is in mmHg/(mL/min), equivalently mmHg·min/mL. The relation assumes stable formation, no artificial drainage, and approximately linear outflow over the range considered.

Choose illustrative values of downstream pressure 7 mmHg, formation 0.30 mL/min, and resistance 10 mmHg·min/mL. The steady pressure is approximately 7 + 0.30 × 10 = 10 mmHg. If resistance increases to 30 with other assumptions unchanged, the new equilibrium becomes 7 + 0.30 × 30 = 16 mmHg. No change in compliance was required to change that equilibrium.

Compliance, C, describes stored volume per pressure change in mL/mmHg. Around an approximately linear operating region, combining resistance with compliance gives a characteristic response time:

The time constant, τ, is in minutes. If C is assumed constant at 0.50 mL/mmHg and Rout is 30 mmHg·min/mL, τ is 15 minutes. In this idealized first-order model, pressure completes approximately 63% of a step toward a new equilibrium in one time constant. For the change from 10 to 16 mmHg, that gives 10 + 0.63 × 6 ≈ 13.8 mmHg after 15 minutes.

These are constructed system parameters, not normal ranges or expected patient trajectories. Real compliance is nonlinear, routes operate in parallel, and formation, downstream pressure, and vascular volume can change. The model explains why transient behavior and equilibrium require different information.

In patients, infusion-derived resistance depends on reaching interpretable conditions and accounting for background changes or leaks. Martins demonstrated such limitations in clinical measurements, while Tans and Poortvliet's study of 58 adults with hydrocephalus showed that resistance and pressure–volume estimates differed with method and model. Neither quantity is a universal diagnosis or stand-alone treatment selector. 11, 12

61.10 Solute Clearance Is Not Synonymous with Fluid Circulation

Clearing a solute requires movement out of the compartment being considered, followed by disposal or transfer elsewhere. Movement within that compartment can redistribute concentration without removing any total amount. A fall at one sampling site may therefore reflect mixing or dilution rather than elimination.

Diffusion moves solutes along concentration gradients. Advection transports them with moving fluid. Oscillatory flow can enhance dispersion, and membrane transport, cellular uptake, metabolism, and blood–brain barrier exchange can contribute differently for different molecules. Water turnover cannot automatically be assigned as the clearance rate of every solute.

The glymphatic framework proposes organized CSF–interstitial exchange involving perivascular routes and astrocytic water handling. Iliff and colleagues' mouse tracer experiments supported paravascular entry and clearance pathways and an influence of aquaporin-4 in their preparation. These findings stimulated important mechanistic work but did not measure a complete adult-human clearance circuit. 13

Smith and colleagues subsequently reported rodent experiments in which local parenchymal solute movement was consistent with diffusion and was not dependent on aquaporin-4 in the manner predicted. The findings illustrate that methods, spatial scale, and the distinction between perivascular transport and transport through tissue matter to interpretation. 14

Neither result justifies treating all brain transport as diffusion alone or as a single established bulk-flow loop. Tracer size, injection conditions, anesthesia, vascular state, and the interval measured can influence what is observed. Evidence that one pathway exists does not determine its quantitative dominance for every solute or condition.

A useful clearance definition is the equivalent volume of fluid cleared of a specified solute per unit time. If concentration is 2 µmol/mL and a defined pathway removes 0.10 µmol/min, its clearance is 0.10 ÷ 2 = 0.05 mL/min, under suitable steady-state sampling assumptions. This constructed solute clearance is not necessarily the physical water flow through the route.

The physiology therefore requires molecule-specific and compartment-specific questions. A general claim that an intervention improves brain “washing” cannot substitute for demonstrating what was transported, where it went, and whether the patient benefited.

61.11 Hydrocephalus Reflects Altered Flow, Storage, and Tissue Mechanics

Hydrocephalus describes pathological ventricular enlargement associated with disturbed CSF physiology. The resulting geometry depends on the distribution of flow restrictions, pressure gradients, tissue properties, and duration. Ventricular size does not specify a unique pressure or formation rate.

An obstruction within ventricular pathways can create upstream accumulation and a pressure difference across the restricted segment. In a communicating pattern, major ventricular connections remain open, but resistance elsewhere, altered downstream pressures, or impaired craniospinal accommodation can still disturb regulation. These categories describe useful patterns without identifying every mechanism.

Hemorrhage and inflammation can affect several parts of the system simultaneously: physical obstruction, outflow interfaces, epithelial transport, and tissue mechanics. Karimy and colleagues demonstrated inflammatory signaling associated with increased choroid secretion in a rat model of posthemorrhagic hydrocephalus. This establishes a possible secretory contribution in that preparation, not that all human posthemorrhagic hydrocephalus is primarily hypersecretion or that an experimental inhibitor improves clinical outcomes. 15

Progressive ventricular expansion can alter surrounding tissue geometry and fluid exchange. Pressure differences across the ventricular wall and extracellular pathways can contribute to movement into adjacent tissue. Detailed edema mechanisms and osmotic gradients belong to Chapter 62; the immediate lesson is that ventricular fluid and parenchymal water are mechanically connected.

The phrase normal-pressure hydrocephalus should not be read as proof of normal dynamics at all times. A mean pressure measured during one interval can coexist with abnormal pulsatility, impaired outflow, altered reserve, or a different chronic mechanical operating point. Conversely, ventricular enlargement can accompany loss of tissue volume without representing the same disorder of CSF regulation.

Age and timescale further modify the picture. A growing skull and developing brain do not have the same boundaries or transport environment as an adult with an intact rigid skull. Slow adaptation can permit considerable geometric change without reproducing the pressure response to an acute obstruction.

Hydrocephalus must therefore be interpreted through anatomy, trajectory, examination, and appropriate dynamic evidence, rather than ventricular size or one pressure value alone.

61.12 Drainage Creates a New Boundary Condition

An external drain or implanted shunt adds an outflow pathway. Its behavior depends on proximal pressure, outlet or distal pressure, valve characteristics, catheter resistance, hydrostatic height, and whether the pathway remains patent. The system after drainage begins is mechanically different from the undrained system.

Lowering an effective outlet pressure can increase flow if the remaining conditions permit it. As volume leaves, however, proximal pressure and ventricular geometry can change, modifying the driving gradient. A catheter may encounter obstruction or collapse, while posture can create siphoning forces in a long fluid column. A valve's nominal opening pressure is therefore not a complete description of flow.

Czosnyka and colleagues' laboratory testing of several shunt designs demonstrated dependence on catheter configuration, outlet pressure, pulsation, and other conditions. Those historical devices do not represent every current shunt, but the experiments establish why a setting or device label cannot replace knowledge of the pressure–flow relationship. 16

Consider an illustrative steady-state accounting in which net CSF formation is assumed to be 18 mL/h, native outflow 12 mL/h, and external drainage 6 mL/h. Total output is 12 + 6 = 18 mL/h, so volume is stable under those assumptions. Measuring the 6 mL/h drain output would underestimate formation if it were mistaken for the entire output.

If drainage subsequently falls, possible explanations include a smaller driving gradient, greater native outflow, altered formation, or obstruction. Output alone does not distinguish them. Likewise, a pressure measured while fluid is being drained reflects the changed system and the measurement configuration, rather than necessarily reproducing undrained pressure.

The interpretation must therefore connect verified pressure, output, patient position, patency, imaging, and neurological trajectory. Changing a drain configuration is a clinical intervention with consequences for volume and gradients; these physiological examples do not specify adjustment procedures or removal targets.

CSF regulation is ultimately the behavior of secretion, connected storage, exchange, and outflow together. Treating one measured stream as the entire system loses the mechanisms needed to understand the response.

References

1. Brightman MW, Reese TS. Junctions between intimately apposed cell membranes in the vertebrate brain. J Cell Biol. 1969;40:648–677. Original ultrastructural study.

2. Oernbo EK, Steffensen AB, Razzaghi Khamesi P, et al. Membrane transporters control cerebrospinal fluid formation independently of conventional osmosis to modulate intracranial pressure. Fluids Barriers CNS. 2022;19:65. Original experimental study.

3. Pollay M, Hisey B, Reynolds E, et al. Choroid plexus Na⁺/K⁺-activated adenosine triphosphatase and cerebrospinal fluid formation. Neurosurgery. 1985;17:768–772. Original experimental study.

4. Steffensen AB, Oernbo EK, Stoica A, et al. Cotransporter-mediated water transport underlying cerebrospinal fluid formation. Nat Commun. 2018;9:2167. Original experimental study.

5. Gregoriades JMC, Madaris A, Alvarez FJ, Alvarez-Leefmans FJ. Genetic and pharmacological inactivation of apical Na⁺-K⁺-2Cl⁻ cotransporter 1 in choroid plexus epithelial cells reveals the physiological function of the cotransporter. Am J Physiol Cell Physiol. 2019;316:C525–C544. Original experimental study.

6. Xu H, Fame RM, Sadegh C, et al. Choroid plexus NKCC1 mediates cerebrospinal fluid clearance during mouse early postnatal development. Nat Commun. 2021;12:447. Original developmental study.

7. Jensen DB, Toft-Bertelsen TL, Barbuskaite D, et al. The Na⁺,K⁺,2Cl⁻ cotransporter, not aquaporin 1, sustains cerebrospinal fluid secretion while controlling brain K⁺ homeostasis. Adv Sci. 2025;12:2409120. Original experimental study.

8. Dreha-Kulaczewski S, Joseph AA, Merboldt KD, et al. Inspiration is the major regulator of human CSF flow. J Neurosci. 2015;35:2485–2491. Original MRI study.

9. Ringstad G, Eide PK. Cerebrospinal fluid tracer efflux to parasagittal dura in humans. Nat Commun. 2020;11:354. Original human tracer study.

10. Albayram M, Richmond SB, Yagmurlu K, et al. Meningeal lymphatic architecture and drainage dynamics surrounding the human middle meningeal artery. iScience. 2025;28:113693. Original human imaging and tissue study.

11. Martins AN. Resistance to drainage of cerebrospinal fluid: clinical measurement and significance. J Neurol Neurosurg Psychiatry. 1973;36:313–318. Original clinical study.

12. Tans JT, Poortvliet DC. CSF outflow resistance and pressure-volume index determined by steady-state and bolus infusions. Clin Neurol Neurosurg. 1985;87:159–165. Original clinical study.

13. Iliff JJ, Wang M, Liao Y, et al. A paravascular pathway facilitates CSF flow through the brain parenchyma and the clearance of interstitial solutes, including amyloid β. Sci Transl Med. 2012;4:147ra111. Original mouse tracer study.

14. Smith AJ, Yao X, Dix JA, Jin BJ, Verkman AS. Test of the glymphatic hypothesis demonstrates diffusive and aquaporin-4-independent solute transport in rodent brain parenchyma. eLife. 2017;6:e27679. Original experimental study.

15. Karimy JK, Zhang J, Kurland DB, et al. Inflammation-dependent cerebrospinal fluid hypersecretion by the choroid plexus epithelium in posthemorrhagic hydrocephalus. Nat Med. 2017;23:997–1003. Original experimental study.

16. Czosnyka M, Czosnyka Z, Whitehouse H, Pickard JD. Hydrodynamic properties of hydrocephalus shunts: United Kingdom Shunt Evaluation Laboratory. J Neurol Neurosurg Psychiatry. 1997;62:43–50. Original laboratory study.