Altered brain fluid dynamics in spontaneous intracranial hypotension

Abstract

Background

This prospective observational study explored alterations in brain fluid dynamics in patients with spontaneous intracranial hypotension (SIH) and confirmed spinal cerebrospinal fluid (CSF) leakage, specifically addressing CSF clearance to blood, glymphatic influx and measures of CSF flow.

Methods

A cohort of SIH patients with verified spinal CSF leaks was compared with an age- and sex-matched reference group having no CSF disturbance or neurological disorders. Prior to repair of CSF leakage, CSF clearance was quantified using population pharmacokinetics, glymphatic influx was assessed by intrathecal contrast-enhanced magnetic resonance imaging (MRI), and CSF flow patterns were measured using either phase-contrast MRI or multi-phase analysis of CSF tracer transport within the subarachnoid spaces.

Results

The study included eight SIH cases and nine reference subjects. SIH was accompanied with a greater CSF clearance, severely reduced tracer enrichment in the subarachnoid spaces and impaired glymphatic influx throughout the brain, after the patients had been upright. However, with patients in the supine position during MRI scanning, times to first enrichment of tracer in cisterna magna or perivascular subarachnoid spaces were not affected, with no changes in CSF flow through the Sylvian aqueduct, indicating that ventricular CSF production remained unaffected.

Conclusions

SIH caused by spinal CSF leakage is accompanied with altered brain fluid dynamics, here illustrated by accelerated CSF clearance to blood and reduced glymphatic influx, likely due to reduced CSF volume load to the intracranial compartment in the upright position.

Background

Spontaneous intracranial hypotension (SIH) is a serious neurological disorder caused by cerebrospinal fluid (CSF) leakage, typically from a spinal source, although it can occasionally originate from a cranial site [1–3]. Its incidence is estimated to approximately 3.8 per 100,000 people [4, 5], and can manifest with a wide range of symptoms, from mild headaches to cognitive deficits and even dementia [3, 6–9], and coma [10]. The symptom burden can significantly affect quality of life and work capacity, with some individuals unable to continue working [11, 12]. While repairing the CSF leak usually alleviates symptoms, some patients may experience persistent issues, raising concerns about whether the CSF leak has been fully repaired, and whether SIH may develop in the absence of any CSF leakage [13]. In the latter category, various pathophysiological events including large thecal sac compliance were proposed to cause an abnormal orthostatic cranial-to-spinal CSF shift [13].

The mechanisms behind the clinical impairments associated with SIH remains to be fully understood. CSF leakage-associated brain sagging and CSF flow alterations with abnormal spinal cord motion may cause increased strain on neuronal structures [14, 15]. Additionally, alterations in CSF dynamics may change over time from acute to chronic stages; lumbar CSF dynamics assessed by lumbar infusion testing were found to be altered only in patients with acute, not chronic, symptoms [16]. The diagnosis of SIH also implies low intracranial pressure (ICP), though ICP is rarely measured, and not recommended as standard clinical routine [8]. Moreover, measuring lumbar opening CSF pressure has very limited value in SIH [9, 17].

The objective of this study was to gain a better understanding of how SIH affects brain fluid dynamics more broadly. In cases with chronic SIH and confirmed spinal CSF leakage, we characterized the CSF clearance and explored the potential effects on glymphatic influx and various measures of CSF flow.

Methods

Study population

The study includes two groups of patients consecutively recruited within the Department of Neurosurgery: a cohort with spontaneous intracranial hypotension (SIH) and a reference (REF) cohort. For both the SIH and REF cohorts, their symptoms and complaints were recorded prospectively.

SIH Cohort

Inclusion criteria for the SIH cohort: (i) Intrathecal contrast-enhanced magnetic resonance imaging (MRI), from here referred to as gMRI (glymphatic MRI), was performed as part of the diagnostic workup of SIH of chronic duration (> 3 months). This was done to identify site of CSF leakage along with other imaging modalities. (ii) A spinal CSF leakage site was identified. Defining the subtypes of spinal leakage was beyond the scope of this study [18].iii) No prior treatment for SIH.

Exclusion criteria for the SIH cohort: i) A primary tentative diagnosis of SIH, but without any identified CSF leakage. ii). SIH patients with confirmed CSF leakage who did not undergo gMRI. Since the majority of SIH patients in our institution do not undergo gMRI, the present study population represents a subset of the full SIH cohort.

REF cohort

Inclusion criteria for the REF cohort: (i) Individuals at comparable age and sex as SIH subjects who underwent the same gMRI protocol as the SIH patients; making the SIH and REF subjects age- and sex-matched. (ii) The primary clinical indication for gMRI was a symptom profile comparable to the SIH patients with a primary tentative diagnosis of an arachnoid cyst or suspected increased intracranial pressure (ICP). (iii) The diagnostic work-up revealed no evidence of obstructed CSF flow to the tentative cyst and no evidence of increased ICP assessed by overnight ICP monitoring, and no other CSF disturbances or neurological conditions.

Exclusion criteria for the REF cohort: (i) SIH was the primary suspected diagnosis even though no CSF leakage was identified. (ii) Other CSF disturbances or neurological disease were identified.

Overall, we consider the REF participants as close to healthy as possible. Intrathecal gadobutrol is given off-label; therefore, we did not include healthy volunteers.

CSF clearance

To measure the CSF clearance, we used a previously described method [19, 20]. The MRI contrast agent gadobutrol was administered intrathecally as a CSF tracer at a dose of 0.50 mmol (0.50 ml of 1.0 mmol/ml gadobutrol; Gadovist, Bayer Pharma AG, Berlin, Germany), both to quantify CSF clearance and visualize glymphatic influx and CSF transport. Plasma samples were obtained at least for every MRI acquisition. The measured gadolinium concentrations in plasma were recalculated to gadobutrol concentrations, and a previously developed population pharmacokinetic model was used to determine individual pharmacokinetic parameters for intrathecal gadobutrol [20]. A two-compartment model with first-order elimination from the central compartment was developed using a non-parametric adaptive grid approach, implemented in P metrics package for R. Posterior individual parameter values and predicted concentrations were obtained from the final population pharmacokinetic model based on the complete dataset.

Predictions were generated at one-minute intervals after administration, and the following pharmacokinetic variables of CSF clearance were included: (i) Absorption half-life (T_{1/2, abs}): The time it takes for half the amount of gadobutrol in the CSF to be cleared into the bloodstream, serving as a surrogate marker for the CSF clearance. (ii) Time to maximum concentration (T_max) in plasma: Derived directly from individual predictions. (iii) Lag-time of absorption (T_lag): The model-estimated time for the tracer to reach the site of clearance in the CSF, with a longer T_lag indicating that the tracer remains in the CSF longer or that the clearance process is delayed. (iv) Maximum concentration (C_max): Also derived from individual predictions. (v) Area under the concentration-time curve from zero to infinity (AUC_0−∞): A measure of systemic exposure to gadobutrol. We have previously demonstrated dose linearity within the relevant range (0.1 to 0.05 mmol) [20].

Magnetic resonance imaging (MRI)

All patients underwent magnetic resonance imaging using a 3 Tesla Philips Ingenia MRI scanner (Philips Medical Systems, Best, The Netherlands) equipped with a 32-channel head coil. The patients were in the supine position during MRI acquisitions and for 2–3 h after spinal puncture and then was allowed to walk around.

Glymphatic MRI (gMRI)

Standardized T1-weighted MRI scans were performed before and 0–1 h, 3.5–7 h, about 24 h, about 48 h and 4 weeks after the intrathecal administration of gadobutrol, maintaining consistent imaging protocol settings across all time points. Sagittal 3D T1-weighted gradient echo scans were used, with the following imaging parameters: repetition time = “shortest” (typically 5.1 ms), echo time = “shortest” (typically 2.3 ms), flip angle = 8 degrees, field of view = 256 × 256 mm, and matrix = 256 × 256 pixels (reconstructed 512 × 512). A total of 184 contiguous slices, each 1 mm thick, were acquired and automatically reconstructed to 368 slices with 0.5 mm thickness.

Gadobutrol shortens the T1 relaxation time of water, increasing T1 signal intensity in the grayscale images, allowing for a semi-quantitative assessment of tracer distribution. Post-processing was performed using FreeSurfer software (version 6.0; http://surfer.nmr.mgh.harvard.edu/), which enabled the calculation of the percentage change in normalized T1 signal units, as well as brain volumes. FreeSurfer allows for segmentation, parcellation, and alignment of longitudinal data, along with the computation of T1 signal changes induced by the CSF tracer, as previously reported [21].

We normalized grayscale changes between scans by dividing the T1 signal units at each time point by those from a reference region of interest (ROI) placed in fat tissue located in the posterior orbit. This ratio represents normalized T1 signal units, adjusting for baseline grayscale variations due to automatic image scaling. CSF tracer enrichment in brain tissue was semi-quantified as the percentage change in normalized T1 signal at different time points compared to pre-contrast injection levels. The subarachnoid space (SAS) was segmented using co-registered T2-weighted volume scans, and parcellation was performed based on the closest and most frequent FreeSurfer parcellation labels.

Phase-contrast MRI utilized for assessing CSF flow at baseline

Phase-contrast MRI (PC-MRI) was done prior to contrast administration. We previously described the method used to measure flow velocity within the Sylvian aqueduct and cranio-cervical junction (CCJ) [22], which is briefly summarized here. For the Sylvian aqueduct, one region of interest (ROI) was placed within the aqueduct’s borders and another in the nearby brain tissue as a reference. Flow velocities were measured at each pixel within the ROI. In a subset of SIH patients, CSF flow volume rates were also measured within the CCJ; for these subjects, a ROI defined the CSF space at level with the CCJ, and another was placed in nearby brain tissue as a reference.

The raw data from PC-MRI were post-processed in MATLAB^® (MathWorks, Natick, United States). Recorded velocities were linearly transformed from pixel values to centimeters per second using velocity encoding. Flow was calculated at each pixel, and reference velocities in the reference ROI were used to identify low signal-to-noise ratios by comparing signal strength to the reference ROI, which highlights the amount of noise. The mean velocity in the reference ROI was also used to detect any bias in the velocity measurements. Aliased velocities (i.e., those exceeding the velocity encoding limit) were corrected using a filter. Mean velocity was determined by averaging recorded velocities for each pixel within the ROI.

CSF volumes in the Sylvian aqueduct and CCJ, both cranially and caudally, were computed by integrating positive and negative volume fluxes over a single cycle. The CSF volumetric net flow rate during one cycle (ml/cycle) was determined by summing the positive and negative CSF fluxes. The net flow rate in mL/min was calculated by multiplying the CSF net flow volume per cardiac cycle by the heart rate (HR). HR was measured during the MRI scan, which lasted 6 min.

Assessment of tracer enrichment in perivascular subarachnoid spaces (PVSAS)

First, tracer enrichment in the PVSAS was evaluated using CSF tracer-enhanced T1-weighted volume scans, analyzed in three orthogonal imaging planes. An experienced neuroradiologist (GR) performed a visual assessment of the tracer distribution patterns within the PVSAS and surrounding regions, following the methodology previously described in greater detail [23].

Statistical methods

Continuous data were presented as mean ± standard deviation (SD) and categorical data as the number of observations. Differences between two independent groups were determined using the independent samples t-test for continuous variables. The associations between continuous variables were assessed using Pearson’s correlation coefficient. The repeated measurements of the percentage change in the normalized T1 signal over time were analyzed using linear mixed models. These models employed maximum likelihood estimation with a subject-specific random intercept and/or distinct residuals varying at time points, considering repeated measurements and model fit. Using the estimated marginal means derived from the statistical model, differences between groups at each time point were tested. Results were presented as mean with a 95% confidence interval (CI). Statistical significance was accepted at the 0.05 level, using 2-tailed P values.

Results

Patient cohort

Patients were enrolled between October 2015 and September 2019, including eight SIH subjects and nine age- and sex-matched reference (REF) individuals. Demographic details and symptom distribution are provided in Table 1. Pre-gMRI symptoms were comparable between the SIH and REF groups. No adverse effects of intrathecal gadobutrol were seen.

Table 1.

Patient material

	SIH		REF	P-value
N	8		9
Gender (F/M)	6/2		7/2	0.89
Age (years)	45.3 ± 16.3		42.2 ± 9.1	0.64
BMI (kg/m 2 )	26.3 ± 5.1		27.7 ± 3.6	0.54
Symptoms prior to MRI
Headache (No/Yes)	1/7		1/8	0.93
Nausea (No/Yes)	8/0		7/2	0.16
Dizziness (No/Yes)	3/5		7/2	0.09
Tinnitus (No/Yes)	4/4		8/1	0.08
Lethargy/Fatigue (No/Yes)	6/2		5/4	0.40
Subjective cognitive impairment (No/Yes)	7/1		5/4	0.15

Data presented as mean ± SD (coefficient of variation given in parenthesis). Abbreviations: BMI: Body mass index. MRI: Magnetic resonance imaging. Continuous variables were compared with independent samples t-test and categorical data with Pearson Chi square test

CSF clearance

Ninety-one blood samples were obtained, with average number of samples being 7.0 ± 1.7 for each patient. The SIH patients (n = 4) exhibited a significantly different CSF clearance profile compared to the REF group (n = 9), characterized by accelerated elimination of the CSF tracer to the bloodstream (Fig. 1A). In SIH patients, the time to 50% clearance of the tracer dose into the blood was significantly shorter, the maximum tracer concentration in the blood was significantly higher, and the time to reach maximum concentration was shorter (Table 2). This confirms that in cases of SIH due to spinal CSF leakage independent of cause, tracer efflux into the bloodstream is faster and more pronounced. However, it is important to note that there was substantial inter-individual variation in the CSF clearance in both the SIH (Fig. 1B) and REF cohorts (Fig. 1C).

Fig. 1.

Accelerated CSF clearance in SIH patients. (A) Predicted plasma concentrations of the CSF tracer (gadobutrol) over time for both the SIH and REF cohorts. (B) and (C) Illustrate the inter-individual variability in CSF clearance among SIH (B) and REF (C) subjects, with black lines indicating the mean concentration for each group, averaged at each time point

Table 2.

Physiological and anatomical comparisons between cases and controls

	SIH		REF	P-value
CSF clearance
T1/2, abs (h)	1.05 ± 0.45		4.53 ± 2.92	0.04
Tmax (h)	2.97 ± 1.36		8.02 ± 4.50	0.05
Cmax (µM)	7.50 ± 3.31		3.36 ± 1.64	0.01
Tlag (h)	0.34 ± 0.39		0.46 ± 0.48	0.67
AUC0−∞ (µM h)	74.71 ± 27.65		68.64 ± 7.23	0.53
Volume measures
Cerebral cortex (mL)	501.96 ± 42.20		512.33 ± 54.20	0.70
Frontal cortex (mL)	196.81 ± 20.66		203.68 ± 23.17	0.53
Temporal cortex (mL)	111.03 ± 12.26		110.39 ± 12.84	0.92
Parietal cortex (mL)	137.39 ± 7.31		140.77 ± 13.34	0.54
Occipital cortex (mL)	56.71 ± 6.35		57.50 ± 7.70	0.82
Cerebellar cortex (mL)	107.94 ± 9.65		105.04 ± 10.89	0.57
4th ventricle (mL)	1.28 ± 0.34		1.58 ± 0.29	0.067
3rd ventricle (mL)	0.93 ± 0.60		0.94 ± 0.32	0.93
Lateral ventricles (mL)	18.04 ± 12.05		19.60 ± 12.04	0.79
Choroid plexus (mL)	1.29 ± 0.35		1.34 ± 0.33	0.74
First time CSF tracer appearance
Cisterna magna (min)	14.50 ± 8.82		10.78 ± 5.22	0.30
PVSAS of anterior cerebral artery (A2; min)	27.00 ± 9.85		39.00 ± 20.58	0.27
PVSAS of anterior cerebral artery (distal)	67.60 ± 45.63		72.40 ± 38.85	0.86
PVSAS of middle cerebral artery (M2; min)	36.20 ± 11.37		33.80 ± 20.00	0.82
PVSAS of middle cerebral artery (M3; min)	80.20 ± 50.45		65.60 ± 43.98	0.64
CSF flow measures
Net antegrade flow Sylvian aqueduct (mL/min)	0.75 ± 1.00		0.26 ± 0.22	0.25
Net upward flow cranio-cervical junction (mL/min)	0.98 ± 0.96		-

PVSAS: Perivascular subarachnoid space. CSF clearance variables: T_{1/2, abs}: Time (minutes) to 50% of tracer dose absorbed to blood (absorption half-life), indicative of CSF tracer clearance to blood. T_max: Time (minutes) to maximum concentration. C_max: Dose-normalized maximum concentration. T_lag: lag-time (minutes) of absorption. AUC_0−∞: Dose-normalized area under curve from zero to infinity. Min: Minutes. Continuous variables were compared with independent samples t-test

CSF tracer enrichment in the subarachnoid space and extravascular (Glymphatic) brain

As shown in Fig. 2A, CSF tracer enrichment in the subarachnoid space (SAS) at 24 h was significantly reduced in SIH patients (n = 8) than REF subjects (n = 9). In the differential (DIFF) image (Fig. 2A, lower row), the blue regions within the SAS indicate the percentage of reduced tracer levels. The time trend plot further highlights the substantial reduction of tracer enrichment in the SAS of SIH patients (Fig. 2B). No significant difference was observed in tracer enrichment within the lateral ventricles (Fig. 2C) or the third ventricle (P > 0.30).

Fig. 2.

Impaired tracer enrichment in subarachnoid spaces and extra-vascular (glymphatic) brain in SIH. (A) Color maps illustrate the percentage change in CSF tracer enrichment within CSF spaces (with the signal change in the brain subtracted) in SIH and REF subjects, measured 24 h after intrathecal (i.th.) tracer injection. The markedly reduced tracer enrichment in subarachnoid spaces is represented in blue in the lower row (DIFF = SIH - REF). The color scale on the right indicates the percentage change in normalized T1 MRI signal units from baseline to 24 h. (B) and (C) Time trend plots show differences in CSF tracer enrichment in the SAS (B) and lateral ventricles (C) between SIH and REF cases. (D) Color maps reveal significantly reduced glymphatic tracer enrichment in SIH subjects at 24 h (DIFF = SIH - REF, lower row). The color bar on the right indicates the percentage change in tracer enrichment at 24 h, with blue in the DIFF images indicating reduced tracer enrichment in SIH cases. (E) Time trend plots demonstrate reduced tracer enrichment in SIH compared to REF subjects within the cerebral cortex and subcortical white matter. These plots show the mean values with error bars representing 95% confidence intervals from linear mixed models. Significant differences between the SIH (red line) and REF (blue line) cohorts are indicated by *P < 0.05, **P < 0.01, and ***P < 0.001

Furthermore, in SIH patients, the reduced tracer enrichment in the SAS was accompanied by severely impaired glymphatic enrichment, as depicted in the differential image (Fig. 2D), where blue areas in the brain parenchyma represent the percentage reduction of tracer enrichment in SIH. The time trend plots further demonstrate significantly reduced tracer enrichment in the cerebral cortex (Fig. 2E) and subcortical white matter (Fig. 2F).

Figure 3 illustrates time trend plots of CSF tracer enrichment in various cortical regions, including the frontal cortex (Fig. 3A), temporal cortex (Fig. 3B), parietal cortex (Fig. 3C), occipital cortex (Fig. 3D), entorhinal cortex (Fig. 3E), parahippocampal cortex (Fig. 3F), hippocampus (Fig. 3G), and cingulate cortex (Fig. 3H). The glymphatic impairment in SIH is widespread, affecting brain regions involved in cognitive function. The widespread cortical glymphatic impairment was not associated with reduced cortical volumes (Table 2).

Fig. 3.

Reduced Tracer Enrichment Over Time in multiple cortical areas of SIH patients. Trend plots show the percentage change in tracer enrichment over time for SIH (n = 8, red line) and REF (n = 9, blue line) subjects across various regions, including: (A) Frontal cortex, (B) Temporal cortex, (C) Parietal cortex, (D) Occipital cortex, (E) Entorhinal cortex, (F) Parahippocampal cortex, (G) Hippocampus, and (H) Cingulate cortex. Significant differences in tracer enrichment over time between SIH and REF cohorts were assessed using mixed model analysis, with significance levels indicated as *P < 0.05, **P < 0.01, and ***P < 0.001

CSF flow imaging patterns

CSF flow patterns assessed with the patient in the supine position revealed no differences between the SIH and REF cohorts. The CSF flow velocities within the Sylvian aqueduct were measured for each pixel (Fig. 4A-B), showing similar CSF volumetric flow rates between SIH (n = 5) and REF (n = 6) subjects (Fig. 4C). In both groups, net antegrade flow (from the third to the fourth ventricle) ranged from 0.08 to 2.43 mL/min. In one SIH patient with a net antegrade aqueductal flow of 2.43 mL/min, it is possible that there was a compensatory increase in ventricular CSF production. This patient also had a choroid plexus volume at the upper end of the range (1.7 mL), though, at the group level, there were no significant differences in choroid plexus volume between SIH and REF subjects (Table 2).

Fig. 4.

Lack of evidence for altered CSF flow measures in SIH patients in the supine position. (A) The region of interest (ROI) within the Sylvian aqueduct is shown in red, while reference ROIs are indicated in blue. The lower right corner displays the pixels included in the analysis, from which flow velocities were calculated for each pixel within the ROI. (B) Gray lines represent the pixel velocities within the aqueduct, with the black line indicating the mean velocity. The colored pixel velocities correspond to the reference ROI, with the black stippled line showing the mean velocity of the reference ROI. (C) The net volumetric flow rate within the Sylvian aqueduct was antegrade in all subjects, with no significant differences between the SIH and REF cases. (D) Strong tracer enrichment within the perivascular subarachnoid space (PVSAS) is illustrated along the middle cerebral artery M2 and M3 branches, characterized by a donut-shaped pattern of tracer enrichment within the PVSAS. There were no significant differences between SIH and REF subjects in the time to first enrichment along anterior cerebral artery (A2 segment; E) or middle cerebral artery (M2 segment; F). Additionally, no differences were observed in the time to first signal change within the cisterna magna (G). Significant differences were assessed using independent samples t-tests

The net volumetric flow rate within the cranio-cervical junction (CCJ) was measured in four SIH patients. In all four, the net flow was upward, meaning the flow direction was from the thecal sac toward the cranial cavity (range 0.29 to 2.39 mL/min; Table 2). Despite CSF leakage, none of these individuals showed evidence of downward net CSF flow (from the cranial to the spinal cavity) when in the supine position.

In the supratentorial SAS, early tracer enrichment was observed within the perivascular SAS (PVSAS) before extending to the outer SAS and brain (Fig. 4D). Tracer enrichment along the anterior cerebral artery (ACA) was seen in 5/6 SIH patients (83.3%) and in all REF subjects (P = 0.30), along the middle cerebral artery (MCA) in all SIH and 5/6 (83.3%) REF subjects (P = 0.30), and along the posterior cerebral artery (PCA) in 2/6 (33.3%) SIH patients and 5/6 (83.3%) REF subjects (P = 0.080, Chi-square test). The time to first enrichment in the PVSAS was comparable between SIH and REF cohorts for both the A2 (Fig. 4E) and M2 (Fig. 4F) segments.

Additionally, the speed of CSF transport within the thecal sac of SIH patients was not delayed. The time from intrathecal injection to first appearance of tracer in the cisterna magna (spinal transit time) was similar between the SIH and REF groups (Fig. 4G).

In summary, these findings, observed with the patient in the supine position, show no evidence of impaired CSF tracer propagation in the SAS, Sylvian aqueduct, or PVSAS of SIH patients.

Discussion

The key findings of this study are that SIH patients with confirmed spinal CSF leakage exhibited an increased CSF clearance, reduced CSF tracer content in subarachnoid spaces, and significantly impaired glymphatic influx. These differences appeared after patients had been allowed to move freely in the upright position.

In the supine position, there were no differences in volumetric flow rates at the Sylvian aqueduct (net downwards) between SIH and REF cohorts, and the net flow rate at foramen magnum was upward in 4/4 SIH patients, despite verified spinal CSF leakage. The discrepancy between the abnormal glymphatic influx and normal CSF flow could be related to the fact that CSF flow was examined with the patient in the supine position, while glymphatic influx was examined after several hours when the individuals had been upright and mobile.

Most studies on SIH focus on methods to detect the site of CSF leakage, which can be highly challenging [24, 25]. Our current diagnostic approach typically involves: (i) Heavily T2-weighted fat saturated myelographic MRI sequences of the total spine. (ii) Dynamic computed tomography (CTM) myelography or dynamic digital subtraction myelography (DSM) in prone and lateral decubitus positions depending on the findings from spine MRI. (iii) Intrathecal contrast-enhanced spinal MRI is used in selected cases to identify CSF leaks. At our institution, this MRI technique was previously used for years to detect CSF leakage in SIH patients. In 2015, we performed cerebral MRI in one SIH patient with late scanning after intrathecal contrast injection (not included in this study) [26], providing the first evidence of the human glymphatic system. The present case series includes all SIH patients with demonstrated spinal CSF leakage who underwent gMRI, making this a representative cohort.

It is not the purpose of this work to advocate methods such as gMRI for imaging CSF leakage, but rather to broadly explore the impact of SIH on brain fluid dynamics. Moreover, we assessed the clinical utility of the CSF clearance in SIH. The technique we used to assess glymphatic function - gMRI (intrathecal contrast-enhanced MRI) - is currently the gold standard for evaluating glymphatic function in humans and has been applied in numerous studies [23, 27–31]. The term “glymphatic” refers specifically to perivascular CSF tracer transport within the brain.

In the age- and sex-matched REF group, there was a clinical indication for performing intrathecal contrast-enhanced MRI, as intrathecal gadobutrol is used off-label and was therefore not administered to healthy individuals. However, no evidence of CSF disturbances or neurological disease was found in the REF group. Notably, the REF subjects were not examined for tentative SIH and do not include individuals with a non-diagnosed SIH. Subjects with a primary suspected SIH but no identified CSF leakage were excluded from the REF cohort. The reason is that CSF leakage may be hard to identify, and we did not want to include subjects with a non-identified CSF leakage. Given that no evidence of CSF disturbance or any neurological disease was identified in these subjects after a thorough work-up, we find it reasonable to consider the REF subjects as close to healthy controls as possible.

The CSF clearance reflects the overall capacity of the system to clear CSF, which we hypothesize is dominated by meningeal-lymphatic clearance [32, 33], and clearance along cranial and spinal nerve roots [34, 35]. From previous studies, we conclude that the CSF tracer is mainly cleared from the spinal canal [20, 36], and less from the upper brain convexity [37]. The test, however, is independent of knowledge about exact clearance routes, and assesses the total CSF clearance over time, which has previously been shown to vary within and between diagnostic groups [20]. Assessing CSF clearance with gadobutrol was not dose-dependent, suggesting very low dose of this exogenous tracer could be applied. In the current study of SIH patients, CSF clearance was accelerated, indicating pathologically increased passage of tracer from the thecal sac, thereby providing physiological confirmation of CSF leakage. While current identification of CSF leaks relies on imaging modalities, CSF clearance could potentially be used in the clinic in the diagnostic work-up of a leak in conjunction with intrathecal enhanced CT- or MRI. It would require 3–4 blood samples a few hours after intrathecal contrast injection. In this regard, CT contrast agents have many features in common with MRI contrast agents with concern to molecular size, hydrophilicity, and their limited crossing of the intact blood-brain-barrier. As such, utilizing a CT contrast agent instead might have large potential for widespread use in assessment of CSF clearance [38]. This could be assessed prospectively in further studies and would not be limited by the off-label use of intrathecal MRI contrast agents.

An alternative to identifying accelerated CSF clearance in patients with non-identified leaks, is imaging of early CSF tracer enrichment in kidneys in conjunction with radionuclide cisternography [38]. This procedure is, however, in limited clinical use, and does not quantify CSF clearance to blood directly. Another strategy might be positron emission tomography (PET) which may identify whether a CSF leakage is present [39]. Even though CSF clearance was accelerated in SIH patients, there also was large inter-individual variation both among the SIH and REF subjects. At present, we have no good explanation for this variability and are left with speculations why it differs so much. Genetic factors may be involved, as well as differences in meningeal lymphatic clearance function. However, further research is required to clarify this.

A major finding of this study was the significantly reduced tracer enrichment within the SAS. Previous research has shown a close correlation between the amount of tracer in CSF spaces and nearby brain tissue [27, 29]. Therefore, the reduced tracer enrichment in the SAS accounts for the observed decrease in glymphatic tracer influx across cortical and subcortical white matter regions. These findings align with a recent report about impaired glymphatic influx in SIH patients [40]. Impaired glymphatic influx may therefore be secondary to reduced amount of tracer in SAS, and not impaired perivascular transport capacity per se. Notably, glymphatic influx of nutrients and other substances, e.g. trophic factors are necessary for normal brain function. We raise the question of whether impaired glymphatic influx could contribute to the cognitive impairment seen in some SIH patients and in dementia patients [3, 6, 7, 9, 41–43]. Glymphatic impairment has been proposed a common pathway to various neurodegenerative diseases [44]. In this study, only 1/8 with SIH had subjective cognitive challenges, while impaired glymphatic influx was observed in brain regions crucial for cognitive function, such as the entorhinal cortex, parahippocampal cortex, hippocampus, and cingulate cortex. SIH was neither accompanied by reduced brain volume, but others previously reported reduced brain volumes in SIH patients [45]. However, toxic deposits in brain, which may occur secondary to reduced glymphatic clearance, is thought to precede cognitive symptoms by many years, perhaps even decades [46].

The widespread reduction in tracer enrichment across the brain was observed after the patients had been mobilized for several hours. Our patients lie flat for the first 2–3 h; the present imaging findings showed no evidence of impaired CSF tracer passage when they were in the supine position. Thus, when the spinal canal and intracranial compartment was at the same horizontal level, there were no differences in spinal transit time, time to initial tracer appearance in the PVSAS of the A2 and M2 segments, or aqueductal CSF flow between SIH and REF subjects. All SIH cases, as well as the controls, demonstrated net antegrade CSF flow through the Sylvian aqueduct, with flow rates comparable to normal CSF production rates of 0.30–0.40 mL/min. In one case, a flow rate of 2.43 mL/min was observed, potentially indicating compensatory increased ventricular CSF production. Further research is needed to confirm this possibility. The present findings of no differences in net aqueductal CSF flow volumes between SIH patients and references compare with previous observations [47]. The observations of net directed CSF flow in the cranial direction in supine SIH- and REF-patients, is in line with previous observations [22], and resonates with the quick spinal transit times of tracer that was previously demonstrated [48, 49].

It may be questioned whether the net upward CSF flow at the CCJ, as shown by phase-contrast MRI, conflicts with the spinal resorption of CSF tracer observed in CSF clearance assessments. It is important to note that the pathways for water and solute efflux may differ. As such, within the spinal canal, there may be both net water production and solute efflux occurring simultaneously. Water produced within the spinal canal can mix with ventricular CSF within cisterna magna before exiting the central nervous system via cranial nerves, meningeal lymphatics, or across blood vessel walls. There is a continuous and substantial exchange of water between the CSF and blood vessels, both at the brain surface and within the choroid plexus [50]. However, the net direction of water movement along these pathways varies across individuals and disease states and is not yet fully understood. Water appears to have several potential efflux routes from the subarachnoid and ventricular spaces, which may differ from those used by solutes. It has been established that the lumbar CSF opening pressure cannot be used for diagnosing SIH [8]^, [17]. Monitoring ICP from the intracranial compartment in patients with SIH or shunt-related CSF over-drainage, on the other hand, typically reveal marked negative mean ICP scores in the upright position [51]. It may be hypothesized that ICP plays a critical role in CSF transport within the SAS, and that normal ICP is essential for effective CSF passage.

Limitations

We acknowledge some limitations of this study. First, the sample size of SIH cases was small, although the differences between groups were statistically significant. Second, tracer movement in CSF is not identical to CSF flow, although the small, hydrophilic tracer should be considered very suitable to track the CSF pathways. Third, it may also be considered a limitation that a minority of SIH patients had subjective cognitive impairment. Further studies are required to explore the association between glymphatic impairment and cognitive failure in SIH. Fourth, determining net flow rates within the Sylvian aqueduct and CCJ is technically challenging though we consider our approach of quantifying flow at pixel level as methodologically sound [22, 52].

Conclusion

The present results confirm altered brain fluid dynamics in SIH patients with identified spinal CSF leakage. The SIH patients with spinal CSF leakage showed severely reduced CSF tracer enrichment in the SAS and reduced glymphatic influx, likely related to reduced CSF volume and low ICP in the upright position. From the present observations, the CSF clearance emerges as a potential tool in diagnosis of symptomatic CSF leaks, complementing imaging modalities used to detect CSF leakage. Importantly, there was no evidence that CSF flow was impaired in SIH-patients when lying flat, as indicated by unaltered time to CSF tracer enrichment in the cisterna magna or PVSAS and normal aqueductal flow.

Abbreviations

ACA

Anterior cerebral artery

CCJ

Cranio-cervical junction

CSF

Cerebrospinal fluid

CTM

Computed tomography myelography

DSM

Dynamic digital subtraction myelography

ICP

Intracranial pressure

gMRI

Glymphatic magnetic resonance imaging

HR

Heart rate

MCA

Middle cerebral artery

MRI

Magnetic resonance imaging

PCA

Posterior cerebral artery

PVSAS

Perivascular subarachnoid space

REF

References

ROI

Region of interest

SAS

Subarachnoid space

SIH

Spontaneous intracranial hypotension

Author contributions

PKE and G.R. contributed to the conception and design of the study. P.K.E, M.H, A.P., Ø.G., G.L., A.L., E.L., L.M.V, and G.R. contributed to the acquisition and analysis of data. P.K.E, M.H, A.P., E.L., L.M.V, and G.R. contributed to preparing the figures. P.K.E contributed to drafting the text. P.K.E, M.H, A.P., Ø.G., G.L., A.L., E.L., L.M.V, and G.R. contributed to editing the text. All authors approved the manuscript.

Funding

Open access funding provided by University of Oslo (incl Oslo University Hospital). Open access funding provided by University of Oslo (incl Oslo University Hospital). This work was supported by grants from Health South-East, Norway (grants 2020068), and from Department of neurosurgery, Oslo university hospital-Rikshospitalet, Oslo, Norway.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

Per Kristian Eide and Geir Ringstad are shareholders in BrainWideSolutions AS, Oslo, Norway, which is a holder of patents US 11,272,841 and US 12,016,651-B2.

Ethics approval and consent to participate

The study received approval from the following authorities: (i) The Institutional Review Board of Oslo University Hospital (2015/1868) where the study was registered in Oslo University Hospital Research Registry (ePhorte 2015/1868). (ii) The Regional Committee for Medical and Health Research Ethics (REK) of Health Region South-East, Norway (2015/96). (iii) The National Medicines Agency of Norway (15/04932-7). The study adhered to the ethical standards outlined in the Declaration of Helsinki (1975, revised in 1983). All participants provided written and oral informed consent before inclusion. The study design was prospective and observational, and as such sample size calculations and patient randomization were not considered relevant.