Cerebrospinal fluid tracer efflux to parasagittal dura in humans

1The human subarachnoid space is shown to be continuous with the paravascular compartment of all brain regions, enabling for cerebrospinal fluid (CSF) to serve as carrier of solutes from brain metabolism into subarachnoid space. Clearance from CSF to blood was for long considered to be dual; one directly to venous blood through arachnoid granulations, and one via lymphatic channels along exiting nerves at the skull base and from the spine. Lymphatic CSF drainage, and the interplay between the lymphatic- and central nervous system (CNS), gained renewed interest when true lymphatic vessels were shown to reside within the dura mater,. Subsequently, impairment of these meningeal lymphatic pathways has been shown to slow paravascular efflux of amyloid-β from interstitial brain fluid and may possibly represent an aggravating factor in Alzheimer’s disease and age-associated cognitive decline. Neuromodulators from meningeal lymphatics can even possibly modulate brain tissue by their direct release into CSF. A recent animal study was, however, unable to reproduce passage of CSF tracer into dural lymphatic vessels and was thus in line with previous observations indicating that the arachnoid constitutes an impermeable barrier layer between the CSF and dura mater,.

2Our aim was to explore how these diverging findings of CSF tracer efflux from animal studies translate into a human cohort. We carried out multi-phase magnetic resonance imaging (MRI) for up to 48 h in order to follow the movements of the MRI contrast agent gadobutrol, serving as a CSF tracer. Gadobutrol is a non-ionic, highly hydrophilic molecule of relatively small size (604 Da), prevented by the intact brain–blood–barrier from leakage into CNS blood vessels when administered intrathecally. The substance is therefore suitable for tracing molecular efflux from the cranial compartment along extra-vascular pathways.

3The study included 18 patients who were under clinical work-up for possible CSF disorders of various etiologies (Supplementary Table 1).

4After intrathecal injection of the CSF tracer, its movement over time via foramen magnum and underneath the base of the brain toward the upper medial cerebral convexities is illustrated in Supplementary Movie 1.

5Tissue within the parasagittal dura having features as specified above (intermediate or high signal at T2-FLAIR (not free CSF) and at T1-BB (not blood)) was visually identified almost exclusively along the middle to posterior segment of the superior sagittal sinus (Fig. 1; Supplementary Movie 2), corresponding well with the levels where cortical veins enter the sinus.

6Even though there were some variations with regard to prominence, tissue in parasagittal dura with features as here defined was depicted in 18/18 study patients and had the typical appearance of longitudinally interspersed areas of up to several millimeters’ transversal width (Fig. 2a, b). After intrathecal injection of CSF tracer, we found enrichment of parasagittal dura at 6, 24 and 48 h and with peak at 24 h (Fig. 2c–h, Tables 1, 2). At group level, there was a highly significant increase in normalized T1 signal units within the parasagittal dura, though more pronounced in nearby CSF (Fig. 2i).

7As expected, CSF tracer also enhanced in macroscopically visible arachnoid granulations, bulging into the sagittal sinus in scattered locations (Fig. 3). As can be noted, the signal change differed somewhat within the CSF of the subarachnoid space, arachnoid granulations and parasagittal dura, suggesting some level of molecular motion restriction between these structures. Supplementary Movie 3 presents an animation of an arachnoid granulation bulging from parasagittal dura in the superior sagittal sinus.

8The tracer enrichment in parasagittal dura was highly dependent on presence of tracer in adjacent CSF, which was shown by their strong positive correlation at all time points (3, 6, 24 and 48 h) (Fig. 4). At late scans (i.e. 24 and 48 h, respectively), the remnants of tracer in CSF typically accumulated at the high cerebral convexities and in regions adjacent to parasagittal dura (Fig. 1). Tubular structures with any resemblance to dural lymphatic vessels as previously reported could not be discriminated from other enhancing tissue within parasagittal dura, and may be related to the image resolution of our MRI scans being restricted to 1 × 1 × 1 mm voxel size.

9For the visual assessment of tracer enhancement on T1-BB below selected cranial nerve outlets at the skull base, this was detected below the hypoglossal canal in 17/18, the jugular foramen in 17/18, the cribriform plate in 2/18, and the foramen ovale in 0/18 individuals (Fig. 5).

10This study demonstrates for the first time direct efflux of a CSF tracer to parasagittal dura in humans. The findings render for the parasagittal dura to represent a bridging link for CSF-mediated exchange of molecules between brain tissue and dural lymphatic vessels.

11The extent of enhancing tissue within parasagittal dura does not resemble the previously reported appearance of true dural lymphatic vessels, which are tubular-shaped structures of sub-millimeter cross-sectional diameter near the venous sinus wall,,, and also far from the classical MRI appearance of arachnoid granulations bulging into venous sinuses. Our findings compare better with previous anatomical reports demonstrating that parasagittal dura along the middle one-third of the superior sagittal sinus harbors a dense carpet of arachnoid granulations medially, being in close association with an extensive network of intradural channels from 0.02 to 2.0 mm in diameter, coursing and coalescing into lateral lacunae from laterally toward the superior sagittal sinus, (illustrated in Fig. 6). The endothelium of intradural channels was proposed to possibly be the final barrier for the absorption of CSF to venous blood, but passage of CSF over this cell layer has never since been confirmed in vivo. The in vivo-based observations from the current MRI study show that parasagittal dura does not drain CSF only, but also molecules, with high level of dependence on their presence in the adjacent subarachnoid compartment. In light of more recent discoveries showing evidence of true dural lymphatic vessels in the superior sagittal sinus wall,, and the fact that lymphatic vessels as well as blood vessels within the dura lack tight junctions, it seems plausible that molecules within parasagittal dura structures may well reach adjacent lymphatic vessels, and perhaps also vice versa. The parasagittal dura may thus represent a bridging link for CSF-mediated exchange of molecules between brain tissue and dural lymphatic vessels.

12The region with tracer enhancement, as presented here, stretches laterally to a larger extent than what can be anticipated from the previous anatomical descriptions of intradural human arachnoid granulations,, which should be highly suggestive for tracer escape to dura outside arachnoid granulations and into the network of intradural channels. To this end, we note that any CSF or molecular escape to arachnoid granulations whatsoever has lately been questioned, and explained by commonly accepted assumption of the impermeable arachnoid-barrier.

13Tracer enrichment in parasagittal dura is furthermore conflicting evidence against a recent animal study, where basal meningeal lymphatic vessels were proposed being more important than dorsal (parasinusoidal) lymphatic vessels for CSF molecular uptake and transport due to their closer proximity to the subarachnoid space in areas with a much thinner dural layer. Even though we also observed signs of lymphatic drainage through basal neuroforamina, the observations do not strengthen the hypothesis that the importance of molecular drainage toward parasagittal structures is inferior to basal outflow. The very thin, linear structures of tracer enhancement we observed below the scull base, and our image resolution with voxel size 1 mm, did not allow us to quantify the relative importance between parasagittal and basal outflow pathways. While our study corroborates findings from animal studies by showing that CSF molecular outflow through skull base foramina occurs also in humans, the presumed importance of the transnasal lymphatic route requires further exploration in humans. In this study, signs of tracer enhancement below the cribriform plate could be visually confirmed in only 2/18 patients. The olfactory system is indeed of proportionally much lesser size in humans than in rodents, potentially suggesting this molecular clearance route is of less importance, at least relative to clearance toward other cranial nerve outlets at the skull base and the parasagittal dura as corroborated by the current observations.

14The present findings also stand in contradistinction to a recent CSF tracer study of mice reporting no signs of tracer propagation through outside the arachnoid layer, and concluding that CSF outflow occurred predominately along cranial nerves through skull neuroforamina. With concern to clearance rate, another rodent study reported very quick exit of tracer from CSF along perineural routes at the skull base. In our human cohort, tracer was present in CSF at a far longer time span (even after 2 days), and the tail of the remaining tracer bolus at late scans along parasagittal dura (Fig. 1; Supplementary Movie 1) may also indicate a net drift of CSF molecules in this direction, rather than toward the skull base. These obvious discrepancies between observations in mice and humans underline the importance of human translational research.

15It has previously been shown that neurotoxic molecules such as amyloid-β and tau are cleared from brain tissue along paravenous pathways (the glymphatic system) and speculated that molecules also drain further along cerebral veins along the brain convexities into dural lymphatic vessels. A MRI study suggested the same based on peak CSF tracer enhancement in brain and cervical lymph nodes occurring at the same time point. It has, however, been demonstrated in rats that from the total amount of MRI contrast agent injected into the cisterna magna, merely 19% enriched the brain parenchyma, the remainder escaped via other routes. Although the dose, injection site and species differences may be important for this fraction, the present observations of high association between tracer enhancement in CSF and adjacent parasagittal dura suggest that molecules within the CSF primarily drain directly to lymphatic pathways. The extent to which this molecular flux may contribute to brain parenchyma clearance remains to be investigated. In this context, previous findings showing that impairment of meningeal lymphatic function slows paravascular influx of macromolecules into the brain, as well as efflux from the interstitial space, may suggest that impaired lymphatic clearance function may exert its negative effect directly on the accumulation of molecules within the CSF, and thereby further upstream on molecular exit from brain into CSF.

16An obvious limitation of this study is the MR image resolution of 1 mm voxel size. The possible interchange between microscopic contents of the parasagittal dura, such as intradural channels and arachnoid granulations, and dural lymphatic vessels, should be further explored in future studies. In the lack of T1-mapping techniques with image resolution even close to 1 mm, we did not have the opportunity to go beyond semi-quantitative measures of tracer enrichment. Ideally, we could also have used an extra-cranial device as reference for normalization of signal units, as the ocular bulbs may be affected by some motion artifacts during scans of several minutes. Moreover, the long duration of molecular clearance from human CSF, and the high demand on patients and resources associated with multi-phase MR imaging over 48 h, has limited the number of scans we were able to perform in single patients through this time span. Therefore, the time point for peak tracer enhancement in human parasagittal dura must be considered to represent a rough estimate.

17In conclusion, the present study demonstrates in vivo escape of a CSF tracer from the subarachnoid compartment over the arachnoid and into the parasagittal dura along the superior sagittal sinus in humans. The parasagittal dura may thus represent a bridging link for CSF-mediated exchange of molecules between brain tissue and dural lymphatic vessels.

18The study was approved by The Institutional Review Board (2015/1868), Regional Ethics Committee (2015/96) and the National Medicines Agency (15/04932-7). Patients were included after written and oral informed consent. The study was registered in Oslo University Hospital Research Registry (ePhorte 2015/1868).

19The study design was prospective and observational: randomization of patients was not relevant; neither was a priori sample size calculation needed.

20The study included consecutive individuals referred to the Department of neurosurgery, Oslo University Hospital—Rikshospitalet, Oslo, Norway, for various CSF circulation disorders (Supplementary Table 1). Exclusion criteria were: history of hypersensitivity reactions to contrast media agents, history of severe allergy reactions in general, evidence of renal dysfunction (i.e. normal glomerular filtration rate, GFR), age <18 or >80 years, pregnant or breastfeeding women.

21Scans were obtained in a 3-Tesla MRI unit (Philips Ingenia®) with a 32-channel head coil.Whole-brain 3D T2-Fluid Attenuated Inversion Recovery (FLAIR), repetition time (TR)/echo time (TE)/inversion time (TI) = 4800/311/1650 ms, echo train length = 167, flip angle = 90°, 2 averages, 1 × 1 × 1 mm voxel size (isotropic), acquisition time = 5 min and 41 s.Whole-brain 3D T1-Black-blood (BB) scan with TR/TE = 700/35 ms, echo train length = 55, flip angle = 80°, 2 averages, 1 × 1 × 1 mm voxel size (isotropic) and acquisition time 4 min and 54 s.Whole-brain 3D T1-gradient echo (GRE) with TR = shortest (typically 5.1 ms), echo time = shortest (typically 2.3 ms), echo train length = 232, flip angle = 8°, 1 average, 1 × 1 × 1 mm voxel size and acquisition time 6 min and 29 s.

22Whole-brain 3D T2-Fluid Attenuated Inversion Recovery (FLAIR), repetition time (TR)/echo time (TE)/inversion time (TI) = 4800/311/1650 ms, echo train length = 167, flip angle = 90°, 2 averages, 1 × 1 × 1 mm voxel size (isotropic), acquisition time = 5 min and 41 s.

23Whole-brain 3D T1-Black-blood (BB) scan with TR/TE = 700/35 ms, echo train length = 55, flip angle = 80°, 2 averages, 1 × 1 × 1 mm voxel size (isotropic) and acquisition time 4 min and 54 s.

24Whole-brain 3D T1-gradient echo (GRE) with TR = shortest (typically 5.1 ms), echo time = shortest (typically 2.3 ms), echo train length = 232, flip angle = 8°, 1 average, 1 × 1 × 1 mm voxel size and acquisition time 6 min and 29 s.

25We repeated the T1-weighted scans at 3, 6, 24 and 48 h, respectively, after intrathecal injection of 0.5 ml gadobutrol (1 mmol/ml) (Gadovist®, Bayer AB, Sweden). Gadobutrol injection was preceded by injection of 3 ml of 270 mg I/ml iodixanol (Visipaque®, GE Healthcare, Norway) to ensure correct needle position within the intrathecal compartment. Patients were kept in the supine position with a pillow under their heads until the 6 h scan was completed and were then allowed to move freely.

26Sagittal T2-fluid FLAIR, as well as T1-BB scans from all time points, were reformatted into coronal and axial 1-mm slices perpendicular and parallel to the plane defined by a line between the anterior and posterior commissures (AC–PC plane), respectively. Identically positioned coronal slices were used to identify parasagittal dura with tissue characteristics remaining unsuppressed (intermediate or high signal) at both T2-FLAIR and T1-BB. T2-FLAIR has the ability to suppress free (unbound) CSF within the ventricular compartment and subarachnoid space. T1-BB images are tuned to darken the contents of blood vessels, even when containing a contrast agent. Thus, in areas with moving protons, such as in arteries and veins, the T1-BB signal is nulled. Around the high brain convexities, there are little motions within the CSF, and therefore the signal is not black (nulled) as in the blood vessels, rendering for a more intermediate CSF-signal.

27In areas of parasagittal dura with these features, regions of interest (ROIs) were placed in identical locations at all time points and well within the outer perimeter of parasagittal dura at T1-BB images for robust measurement of T1 signal change as a sign of tracer enrichment. In addition, a ROI was placed within adjacent subarachnoid space to assess T1 signal change in CSF. The CSF tracer enrichment was measured as change in T1-BB signal units, and was normalized against the reference (i.e. the vitreous body of the ocular bulb) to correct for any baseline shift of image grayscale between time points (Supplementary Fig. 1). We also visually assessed any presence of tracer enhancement on T1-BB images below selected cranial nerve outlets at the skull base and dichotomized this as yes/no whether such enhancement was found present at any time point, or not, respectively.

283D T1-GRE scans were used for segmentation and co-registration rendering for 3D representations (using 3D Slicer version 4 and SPM 12, both open source software) illustrating the relations between the brain surface and CSF tracer enhancement. T1-GRE was, however, unsuitable for precise ROI-placement in parasagittal dura, since the border against the enhancing CSF compartment became effaced at late scans (Supplementary Fig. 2).

29Statistical analyses were performed using the SPSS software version 22 (IBM Corporation, Armonk, NY). Differences between continuous data were determined using linear mixed models with a random intercept. Correlations were determined by Pearson correlation coefficient. Statistical significance was accepted at the 0.05 level (two-tailed).

30Further information on research design is available in the Nature Research Reporting Summary linked to this article.