Abstract
Background
Spontaneous skull base cerebrospinal fluid leaks (CSFLs) are associated with increased intracranial pressure in idiopathic intracranial hypertension (IIH) and hypothesized to relate to skull base erosions due to increased CSF pressure. Given the increasing recognition of internal jugular venous stenosis (IJVS) as a cause of intracranial hypertension (IH), we evaluated the relationship between spinal CSFL and venous causes of IH.
Methods
The spinal CSFL database at a single institution was assessed to identify 12 consecutive spontaneous, non-traumatic spinal CSFL patients with CTV data. Exclusion criteria included documented IIH and iatrogenic CSFL. Demographics, clinical parameters, imaging characteristics, and IJV manometry results were recorded. Internal jugular venous stenosis was graded as: none (0–10%), mild (10–50%), moderate (50–80%), severe (>80–99%), and occluded (100%). Twelve consecutive patients who presented with cerebrovascular accidents without CSFL, matched by age and sex, were similarly analyzed as a control group. STROBE guidelines were used in reporting results.
Results
All CSFL patients had IJVS (83.3% bilateral, 33.3% severe) compared to 41.7% of the control group (33.3% bilateral, 16.7% severe-occluded); p = 0.04. All CSFL patients with available venogram manometry data had at least unilateral IJV gradients. Most patients presented with modified Rankin score (mRS) of 1 (66.7%), but in those with higher mRS, medical and/or surgical interventions were associated with decreased morbidity.
Conclusion
Spontaneous spinal CSFL was associated with IJVS in patients not meeting IIH criteria. Persistently high CSF pressure resulting in CSFL may cause opening pressure to be falsely normal or low. Internal jugular venous stenosis may be a viable target in recurrent CSFL management and improve morbidity.
Introduction
While the mechanisms explaining development of spontaneous and/or recurrent cerebrospinal fluid (CSF) leak are not well understood through the literature, certain outflow pathway obstructions have been implicated. 1 Cerebrospinal fluid has been shown to be resorbed from the subarachnoid space through arachnoid herniation into venous sinuses with intracellular vacuoles serving to transport fluid proportionate to the differential pressure between the CSF and veins. 2 Additionally, CSF also drains via the lymphatic system in the skull base and spine. 3 While normal CSF pressure drives resorption from the subarachnoid compartment into the venous system, maintenance of an elevated intracranial pressure contributes to a dysfunctional hydrostatic stress upon arachnoid granulations and dura, as well as the balance of venous and lymphatic resorption of fluid. 2 The lymphatic proportion of CSF drainage increases substantially to compensate for venous outflow dysfunction, but is unable to do so indefinitely and is ultimately overwhelmed. 4 The continued stress on the dura due to the elevated CSF pressure leads to its weakening and eventual rupture.
Idiopathic intracranial hypertension (IIH) (or pseudotumor cerebri) can result from reduced cranial outflow of fluid relative to CSF production when a mass or hydrocephalus are not the contributing factors to the pressure derangement. These disorders predominantly affect obese women of child-bearing age and manifest with frequent headaches. 5 While certain intracranial sites of the venous outflow system have targeted treatment methods described, others remain less explored.1,6–8 Offering complimentary extracranial treatment strategies is important to this patient population, as they may not only experience disabling symptoms but might also undergo unnecessary costly tests and surgeries. 9 These alternatives may reduce or prevent morbidity of intracranial hypertension (IH) and CSF leak that includes symptoms of dizziness, memory decline/difficulty, tinnitus, high-frequency hearing decline, eye bloating, diplopia (sixth nerve palsies), blurred vision, visual field defects (papilledema), neck discomfort, and sleep difficulty. 10
Certain patients with select anatomic and physiologic outflow disturbance demonstrable upon dynamic imaging may be ideal candidates for these developing treatment strategies. While the literature has recognized that the dural venous sinus system and the internal jugular veins (a majority of patients demonstrating right dominant jugular venous drainage) have variability in development, most attention on treatment focuses on the dural venous sinus contribution to outflow disorders on IH, with early data exploring jugular venous stenting.11–14 The craniocervical junction contributes fibroligamentous, osseous, and congenital variability which provide sources of extrinsic impingement that are being targeted for treatment. Not only do dynamic imaging findings support additional contributions by the proximal internal jugular vein (IJV) to outflow dysfunction and elevated intracranial pressure but these can also be demonstrated through catheter angiography, similar to the dural venous sinuses.15,16
While multiple factors contribute to the capacitance of the jugular venous system on imaging, including hydration status, cardiovascular state, head position, and intrathoracic pressure, our hope is to scrutinize the fixed extrinsically compressive anatomic structures that may contribute to IH. 17 Additionally, the internal jugular venous system demonstrates significant size, angulation, and symmetry variability that could potentially confer susceptibility or resistance to outflow disturbance. Through the following cases we hope to establish support for these findings to be included in the search pattern of the neuroradiologist, increase clinician awareness about site contribution to IH, as well as guide interventionalists in treatment options for CSF leaks.
Methods
In accordance with the local IRB (NA_00029413), the CSF Leak Database at our institution was queried to identify consecutive patients with spontaneous and/or recurrent nontraumatic CSF leaks who subsequently underwent venous CT imaging of the skull base and neck between 2016 and 2021. The CSF leak database contains all patients with high- or low-pressure CSF disorders seen at our institution's Hydrodynamics Center, including IIH, CSF leaks, and hydrocephalus, of which a consecutive series of patients referred for blood patch treatment of spontaneous intracranial hypotension (SIH) were included in our study. Given the relative paucity of patients meeting these criteria, study size was dictated by the number of patients within this time period. Exclusion criteria included documented IIH, as the diagnosis entails the exclusion of other causes of IH, or iatrogenic causes of CSF leak. Demographics, comorbidities, and clinical parameters were identified and noted. Internal jugular venous stenosis was quantified as the narrowest point on CTV relative to the cross-sectional area at the jugular foramen and graded as: none (0–10%), mild (10–50%), moderate (50–80%), severe (>80–99%), or occluded (100%) and reported by the highest degree present per patient as well as by individual IJV. Myelography was reviewed to identify the presence of dural diverticula. Finally, modified Rankin scores (mRSs) were used to apply a quantitative measure for patients’ morbidity as a result of their disease and treatment courses were followed out to August 2023 as available on the electronic medical record system.
To serve as a control group, 12 consecutive patients presenting with cerebrovascular accidents (CVAs) without CSF leak symptoms who underwent CT angiography (CTA) with sufficient venous phase contrast to evaluate the venous system were analyzed for the same clinical parameters, as well as IJV stenosis. Cerebrovascular accident patients were chosen as the control group given the availability of comparable imaging for IJV stenosis in this population and lack of established association between spontaneous CSF leaks and CVA. They were matched on basis of age and sex. Given the small sample sizes, Fisher's exact test was performed with grouped IJV stenosis categories—none/mild and moderate/severe/occluded. The STROBE guidelines for case–control studies were used in reporting the results of the study.
Results
We identified 12 patients from our institution with CSF leaks diagnosed between 2016 and 2021, with ages ranging from 30 to 74 years old (mean age 50.3 years, 50% female, Table 1). Presenting symptoms included headache (100%), positional neck symptoms (75%), brain fog (66.7%), pulsatile tinnitus (41.7%), dizziness (41.7%), and sinus congestion (25%). Headaches were worse with standing (58.3%), bending forward (25%), and/or Valsalva maneuver (16.7%). Most prevalent comorbidities included migraine with aura (33.3%), hypertension (25%), and Chiari malformation type 1 (25%). Computed tomography myelography demonstrated dural diverticula (83.3%) and/or CSF leaks (33.3%) in all patients. Further, all had some degree of IJV stenosis, whether mild (25%), moderate (41.7%), or severe (33.3%), with bilateral IJV involvement in 83.3%. Of the 12 patients, seven had venous manometry data available, with two patients demonstrating unilateral IJV gradients, and the other five patients demonstrating bilateral IJV gradients, with a mean gradient of 2.8 mmHg. Opening pressure (OP) was available for 9 of 12 cerebrospinal fluid leak (CSFL) patients and was normal in 66.7% (19.1 ± 8.7 mmHg). In comparison, in patients who underwent CTA for CVA (mean age 50.8 years, 42% female, Table 2) only 16.7% presented with the queried presenting symptoms seen in CSFL—one patient presented with headache and aura, while another presented with dizziness. Additionally, there was a much lower prevalence of IJV stenosis (41.7% total, 16.7% mild, 8.3% moderate, 8.3% severe, 8.3% occluded, 33.3% bilateral; Table 2). Fisher's exact test performed with grouped IJV stenosis categories yielded a statistically significant p-value of 0.04.
CSF leak patient characteristics and outcomes.
Control non-CSF leak patient characteristics.
With the exception of one patient whose symptoms resolved without intervention, all patients underwent an epidural blood patch procedure (ranging between 1 and 10 epidural blood patches), with 58.3% requiring multiple. Two patients in the series underwent IJV release procedures (styloidectomy or tuberculectomy), one of whom is described in a case report by Primiani et al. 9 We present below two unique cases from our series, one of whom required multiple repeated blood patches and may benefit from IJV release surgery, and one of whom did not require any intervention and spontaneously improved on their own.
Case 1
A patient in their seventh decade of life with a past medical history of hypertension presented with subacute worsening of chronic head pressure and right-sided tinnitus and was diagnosed with CSF leak/intracranial hypotension on CT myelogram. The headache was worse on the right, positional, and was associated with pressure behind their eyes, blurry vision, and radiation down their neck. Turning their head to the left or right, sitting, or standing exacerbated their symptoms, while lying down alleviated the headache. Their headache had progressively worsened to the point of causing persistent light-headedness and vertigo when walking, occasionally getting so severe they felt as though they would pass out, causing them to “barely [be able to] take care of [themselves].” They denied any confusion, lethargy, or gait instability. On physical exam, they had mild bilateral dysmetria of their upper and lower extremities, as well as dysdiadochokinesia that typically would localize to the cerebellum, without additional eye movement abnormalities, including papilledema and visual field deficits. CTV demonstrated severe (>95%) narrowing of the patient's right IJV due to compression between the posterior belly of the digastric muscle and transverse process of C1 (Figure 1(a) and (b)). Their left IJV also demonstrated critical stenosis/occlusion at C1–2 from extrinsic compression (Figure 1(c)). Venous manometry demonstrated a right IJV gradient of 4 mmHg and no gradient on the left. Computed tomography angiography did not demonstrate any high-grade intracranial vessel abnormalities. CT post-myelogram prior to their first blood patch demonstrated multilevel perineural cysts of the thoracic and lumbar spine with large foraminal Tarlov's cysts bilaterally at S2, including extrathecal contrast extending from neural foramina at C6–7 and C7–T1, possibly contributing to CSF leak. Opening pressure at this time was 10 mmHg.
CTV demonstrating bilateral severe IJV stenosis. Right IJV extrinsically compressed between (a) transverse process of C1 and (b) posterior belly of the digastric muscle. Left IJV extrinsically compressed at C1–C2 (c).
The patient had transient improvement with epidural blood patches—receiving a total of 10 blood patches on three different occasions. After their third blood patch procedure, which did not provide the same symptomatic relief as their first two, they were started on Diamox for what was thought to be a high-pressure headache presenting as nonpositional posterior skull base pressure. However, they subsequently presented again with a positional headache thought to be associated with low CSF so their Diamox was held, with improvement in their symptoms. As such, the patient was encouraged to maintain close outpatient follow-up to titrate their Diamox regimen in correlation with their headaches, although, as neurology had sufficient clinical suspicion that the patient's symptoms were due to cerebral venous congestion, the patient was informed about C1 tuberculectomy as a possible treatment option as well.
Case 2
A patient in their third decade of life with past medical history of hypertension presented with acute onset neck pain and stiffness associated with posterior positional headaches. The pain intermittently radiated down their shoulders and arms, mostly on the right side, and was relieved by lying down. Exacerbating factors included coughing, sneezing, and Valsalva. They had no trouble swallowing, lower cranial nerve deficits, visual symptoms, or numbness/tingling. They were occasionally dizzy when walking, but had no fine motor deficits, bowel/bladder incontinence/retention, imbalance, or weakness. CT myelogram demonstrated multiple levels of contrast extravasation the thoracic and lumbar spine (right T7–T8 and bilateral L4–L5 and L5–S1 neural foramina) compatible with CSFL as well as multiple tiny perineural sleeve cysts throughout the cervical and thoracic spine. CTV demonstrated mildly diminished caliber of their bilateral IJVs along the tips of the styloid processes, unchanged with neck flexion or extension (Figure 2). Venous manometry demonstrated a left IJV gradient of 1 mmHg and no gradient on the right. As cervical spine MRI showed cerebellar tonsillar ectopia and cervical intramedullary T2 signal hyperintensity, the patient was referred to neurosurgery for styloidectomy. The patient ultimately had resolution of their neck pain and headache four months after its onset despite no procedural intervention. Their neck pain returned two months later, but they elected not to pursue surgical intervention.
CTV showing mild bilateral IJV stenosis. Extrinsic IJV compression by the styloid processes on both the right (a, c) and left (b), unchanged with flexion (d) or extension (e).
Discussion
The jugular outflow system begins as a bulbous dilation at either or both sides at the level of the jugular foramina. Within the distal aspect of these venous conduits above the bulb, valves are present (in 86–93% of individuals) which help to limit transmission of pressure intracranially by closing during the end of diastole. 18 Below the foramen, the jugular vein often drapes over the transverse process of C1 to a variable degree. This interface may serve as contributing factor to outflow obstruction, especially if the osseous impression is atypically large upon the dominant venous outflow tract or if the nature of the interface does not allow for normal capacitance of the venous channel. In some individuals, an abnormally thickened styloid process or abnormally ossified stylohyoid ligament may contribute mass effect upon the jugular vein, carotid, and cranial nerves within the carotid sheath.18,19 As the jugular courses through the neck toward the thorax, its medial border abuts the carotid artery, especially at the carotid bulb which may serve as an extrinsic source of compression. Nearer the thoracic inlet, the sternocleidomastoid overarches the jugular vein as it joins the subclavian vein, yet another opportunity for compression. 18
The Monro–Kelli doctrine describes the limitation of the cranial compartment for changes in volume and associated pressure between the parenchymal, blood, and CSF contributions. 20 While brain parenchyma maintains a relatively static volume contribution, both the absorption of CSF and the outflow of venous blood significantly impact changes in intracranial pressure because of differential volume. Patients with IH typically present with headache related to pressure upon the meninges, nerve roots, and sensitive brain structures. Therefore, diagnosis may be established by lumbar puncture not only demonstrating elevated opening spinal fluid pressure >24 cm H2O but also demonstrating relief of symptoms by relieving pressure through drainage of fluid (especially to pressures below that considered elevated <20–25 cm H2O). 21 However, it is important to note that patients who have sustained elevated CSF pressure to the point of developing dural diverticula and subsequent CSF leak may no longer demonstrate said elevated OPs. Conversely, their OPs may be falsely normal or low, as was the case in 66.7% of our patients in this series. Although only one of our included patients had diagnosed connective tissue disease, this patient population has a higher prevalence of IIH and can present with severe symptoms despite OP in the 15–24 cm H2O range and is more likely to develop CSF leaks after lumbar puncture. This being the case, a major limitation in our retrospective study is that intracranial hypotension was diagnosed clinically with response to CSF leak patching, without demonstration of documented SIH by ICHD-3 criteria—only two of our patients had radiographically visible CSF leakage, and none had documented OPs of less than 6 cm H2O. 22
In patients who have elevated CSF pressure, lumbar puncture temporarily offers relief and serves as part of the diagnostic criteria, but it remains impractical to subject patients to extended or lifelong lumbar punctures to remove volume every time they reestablish intracranial hypertension and develop a headache. Early treatment strategies have included using medical agents, such as carbonic anhydrase inhibitors (e.g., acetazolamide), or inserting a lumbar peritoneal shunt to reduce CSF production and direct excess CSF into a cavity where it can be resorbed without building pressure, respectively. 21 These strategies are somewhat successfully employed, if no direct venous outflow pathway disruption can be identified (i.e., truly idiopathic and/or at the level of the arachnoid granulations). The patient in our first case report is an example of one who intermittently required blood patches and Diamox at different points of their presentation as they ostensibly had both high- and low-pressure headaches throughout their treatment course.
As the larger venous outflow structures lend themselves to both easy recognition of anatomic narrowing, physiologic pressure gradients, and corrective stenting or bypass strategies, the combined conduit scoring system has largely helped physicians recognize and treat transverse and sigmoid sinus stenosis contributions to IH. 23 While this has been a successful method, a largely overlooked complement to this treatment approach includes fixed outflow stenosis within the neck—such as in our patients with IJV stenoses. The majority of our CSF leak patients had IJV stenoses bilaterally and, of those who had manometry, all had at least unilateral IJV pressure gradients. With this clinical pattern identified, it will be key in future prospective studies to demonstrate confirmed SIH by the ICHD-3 criteria on patient presentation, resolution with patching, followed by repeat lumbar puncture demonstrating rebound elevation of CSF pressure along with CTV demonstrating IJV stenosis, and subsequent normalization of CSF pressure again with stenting.
While causation cannot be established with our limited data, the Fisher's exact test comparing the prevalence of IJV stenosis in CSFL patients compared to the control group was statistically significant (p = 0.04), and these findings suggest that larger studies are warranted to assess the association and causality between venous narrowing and spontaneous CSF leaks. Most of our patients presented with mRSs of 1 (66.7%), but of those who had a higher mRS, appropriate interventions, both surgical and medical, led to decrease in morbidity. Firstly, in the patient who underwent C1 tuberculectomy as described in Primiani et al., mRS decreased from 2 to 1 with the procedure. 9 Secondly, in the patient in our first case who underwent blood patches with close titration of their diamox regimen, mRS decreased from 3 to 2. Although CSFL was visualized in the lower cervical and thoracic levels, venous compression at the C1 level not only increases intracranial CSF pressure but is translated distally given the continuity of the CSF space and would be expected to result in generalized CSF column pressurization. While the small size and heterogeneity of the CSFL population in our institutional database pose a limit on how widely we may extrapolate our hypothesis, it is reasonable to consider that in a patient like this where severe bilateral IJV stenosis has been demonstrated with recurrent CSF leaks requiring multiple blood patches and close follow-up, IJV release surgery may be a beneficial option which could reduce morbidity associated with their fluctuating CSF pressures.
Of note, while we excluded patients with IIH from our study as this entails the exclusion of other causes of IH, it is possible that patients in whom IH of venous etiology was not investigated might be misdiagnosed with IIH, as not all IIH patients undergo CTV with manometry prior to diagnosis, and thus may represent a hidden population that would benefit from further research in this arena as well.
The challenge, then, will be to generate a scoring system or rationale with which to help guide clinicians in their judgment of which patients would be most likely to benefit from IJV release surgery when any degree of IJV stenosis is seen on imaging. In determining presence of IJV stenosis, it also is important to note that stenosis may be positional and therefore CTV performed only in the recumbent position may not be sufficient to reveal IJV stenosis in some patients—for several of our patients, a pressure gradient across the stenosis of greater than 8 cm H2O was only demonstrated in neck rotation, flexion, or extension, with corresponding elevation in intracranial venous pressure. The generation of a database of patients with moderate and severe IJV stenosis, both bilateral and unilateral, who undergo either IJV stenting or styloidectomy or tuberculectomy as indicated for the location of their stenosis is crucial to the development of such a decision-making tool. Aspects of patient history, including recurrent CSF leaks requiring multiple blood patches, as well as headache characteristics—whether only low-pressure or intermittently low- and high-pressure—will be helpful components of the scoring system as well, as seen in our case series.
Conclusion
We present 12 consecutive patients with spontaneous spinal CSF leaks who were clinically imaged with provocative CT venography. All of the patients had substantial IJV narrowing and, of the patients who had venous manometry, at least unilateral IJV pressure gradients, suggesting a strong association between venous narrowing and spinal CSF leaks. There was a statistically significantly lower prevalence of IJV narrowing in a control group of patients without CSF leaks (p = 0.04). Further research will be necessary to explore these findings in a larger population and establish causality, but at minimum, the higher prevalence of venous narrowing compared to a control group suggests that it may be valuable to assess CSF leak patients for premorbid symptoms in the IH spectrum.
Footnotes
Author contributions
FH and CP contributed to the design of the research and collection of data. SM contributed to the collection of data, analysis of the results, and writing of the manuscript, with input/contribution from KF, MA, LG, WS, ML, and FH.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
