Early brain MRI changes following transvenous embolization of cerebrospinal fluid-venous fistulas in spontaneous intracranial hypotension

Abstract

Objective

To evaluate early and mid-term imaging and clinical outcomes following transvenous embolization of cerebrospinal fluid-venous fistulas (CSFVFs) in patients with spontaneous intracranial hypotension (SIH).

Methods

From November 2022 to November 2024, 60 consecutive patients with SIH and confirmed CSFVF underwent transvenous embolization using Onyx. Of these, 40 patients underwent brain MRI pre-treatment, 24 hours post-treatment, and at a 3-month follow-up. The primary outcome was regression of brain MRI abnormalities at 24 hours and 3 months. Secondary outcomes included rates of symptom improvement, predictors of clinical improvement, and complication rates.

Results

The mean patient age was 61 years, and 65% were female. All procedures were technically successful. The median SIH score significantly decreased from 6 pre-treatment to 3.5 at 24 hours (P=0.01) and to 2 at 3 months (P=0.004). Early improvement in SIH score correlated with clinical improvement at 24 hours (P=0.002), which was observed in 77.5% of patients. Pachymeningeal enhancement (87.5%) and venous sinus engorgement (75%) were the most common MRI abnormalities. Both findings regressed in approximately 50% of patients at 24 hours and in 80% of patients at 3 months. At 3 months, 82.5% of patients achieved complete clinical recovery. Rebound post-treatment headaches occurred in 32.5% of patients but resolved within 7 days. The morbidity rate was 0%.

Conclusions

Transvenous embolization of CSFVFs results in early and sustained clinical and imaging improvements in patients with SIH. These findings support the efficacy of this intervention as a primary treatment for CSFVFs.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Transvenous Onyx embolization is an effective and emerging treatment strategy for cerebrospinal fluid-venous fistulas (CSFVFs).

WHAT THIS STUDY ADDS

• Brain MRI findings of spontaneous intracranial hypotension significantly improve as early as 24 hours after CSFVF embolization.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• This study provides new evidence supporting the effectiveness of transvenous Onyx embolization in treating CSFVFs, reinforcing its role as a primary therapeutic option.

Introduction

Cerebrospinal fluid-venous fistulas (CSFVFs) are increasingly recognized as a cause of spontaneous intracranial hypotension (SIH),¹ largely due to technical advancements for localizing such fistulas.^{2 3} SIH significantly impacts patients’ quality of life through a wide range of symptoms, including headache, audiovestibular disturbances, and cognitive impairment, while being associated with high rates of depression, anxiety, and disability.4,6

Transvenous Onyx embolization has recently emerged as a treatment for CSFVF, demonstrating high clinical and radiological efficacy at follow-up.^{7 8} In 2022, after having initiated transvenous CSFVF embolization with Onyx at our center, we observed that most patients experienced symptom improvement within 24 hours post-treatment.^{7 9} To investigate whether these rapid clinical changes were accompanied by early radiological improvement, all consecutive patients underwent a brain MRI assessment 24 hours after embolization, in addition to standard follow-up at 3 months. The purpose of this study was to evaluate changes in symptoms and brain MRI findings during the early period following CSFVF embolization, and to compare their progression with outcomes observed at mid-term follow-up.

Methods

Patient selection

From November 2022 to December 2024, all consecutive patients with SIH and a definite diagnosis of CSFVF were prospectively added to our institutional database (IRB-MTP_2023_01_202201315). Diagnosis of SIH was based on clinical symptoms and brain MRI signs based on the SIH score.¹⁰ The presence of CSFVF was confirmed through the opacification of a vein in the neural foramen and/or paraspinal vein through lateral decubitus dynamic digital subtraction myelography (DSM) or CT myelography. CSFVF embolization was approved at a multidisciplinary CSF dynamics disorders team meeting.

Starting in November 2023, all patients undergoing CSFVF transvenous embolization were monitored with brain MRI at 24 hours post-treatment to detect early changes following leak repair. The first follow-up MRI was performed at 3 months after embolization. MRI was repeated at 6 and 12 months after treatment.

Primary outcome

The primary outcome was the regression of brain MRI signs in the early period (24 hours) and at 3 months after CSFVF embolization, as quantified by the SIH score.

Secondary outcomes

The secondary outcomes included evaluating: (1) clinical symptom improvement percentage at 24 hours and 3 months, (2) the overall complication rate at 3-month follow-up, and (3) predictors of clinical improvement at 24 hours and 3 months.

MRI evaluation

The brain MRI protocol (3 Tesla MAGNETOM Vida, Siemens Healthcare) included the following sequences: three-dimensional (3D) T1-weighted imaging pre- and post-contrast, 3D fluid attenuated inversion recovery (FLAIR), T2 gradient echo (GRE), axial diffusion-weighted imaging (DWI), and susceptibility-weighted imaging (SWI). Initial diagnostic brain MRI was performed within 2 weeks before treatment in all patients, of which 60% were conducted within 5 days before treatment. The initial SIH score was quantified on the MRI before CSFVF embolization. The early post-treatment SIH score was calculated on the 24-hour MRI, and the mid-term post-treatment SIH score on the 3-month follow-up MRI. Pre-treatment MRI findings were compared with those at early and mid-term follow-up.

The SIH score is a quantitative scale focusing on six radiological SIH-related characteristics: venous sinus distention (2 points), pachymeningeal enhancement (2 points), effacement of the suprasellar cistern (SSC) (≤4 mm, 2 points), subdural fluid collections (1 point), effacement of the pre-pontine cistern (PPC) (≤5 mm, 1 point), and reduced mammillopontine cistern (MPC) (≤6.5 mm, 1 point). In addition, we measured the diameter (mm) of the related structures, notably of the posterior aspect of the superior sagittal sinus (SSS) and the dominant transverse venous sinus (TVS), both on the coronal gadolinium-enhanced 3D T1-weighted imaging, as well as the diameter of the SSC, the PPC, and the MPC. An example of the measurement of the cistern and venous structure diameters is provided in the supplemental material (online supplemental figure 1).

Two neuroradiologists, not involved in CSFVF treatment, evaluated the brain MRI. Discrepancies were resolved by a third neuroradiologist who performed the CSFVF embolization. Interobserver agreement was assessed using the kappa (κ) test for the SIH score, SSS diameter, and TVS diameter. These measurements were evaluated before treatment, 24 hours after treatment, and 3 months post-treatment.

As part of the diagnostic algorithm, MRI of the complete spine (3D high-resolution SPACE (Sampling Perfection with Application optimized Contrast using different flip angle Evolution) MRI with and without fat suppression) was performed to rule out spinal longitudinal extradural CSF collection.

Clinical evaluation

All patients underwent comprehensive neurological evaluation before treatment, 24 hours after embolization, and at 3 months. Evaluations were conducted by the neuroradiologist who performed the embolization and by a neurologist who specialized in headaches. The following symptoms were assessed before treatment, at 24 hours, and at 3 months: fatigue, tinnitus (unilateral or bilateral), dizziness, orthostatic headache, non-orthostatic headache, activity-related headache, and Valsalva-related headache. Additionally, patients were evaluated for the presence of chronic headache (≥15 episodes per month for at least 3 months) or episodic headache (≤15 episodes per month).¹¹

After treatment, patients were transferred to the neuroradiology intermediate care unit for a period of 48 hours. The patient was instructed to remain lying down during the first night following treatment, then to stand and walk starting the next day. Additionally, the patient was asked to perform specific activities, such as walking for 10 min, climbing stairs, bending forward to assess activity-related headache, orthostatic headache, and dizziness, as well as coughing for 10 s or performing a Valsalva maneuver to evaluate Valsalva/cough-induced headache. Patients were asked to describe their symptoms and indicate whether they had improved, completely resolved, remained unchanged, or worsened. Symptoms were classified as improved if any symptom showed resolution or improvement (eg, orthostatic headache resolved, but dizziness remained unchanged).

Early-periprocedural (within 48 hours) and mid-to-late (during follow-up) treatment-related complications were recorded. Potential post-treatment rebound headache (PRH) was monitored by clinical evaluation and was defined as a change in headache features, notably, by the headache becoming non-orthostatic or worsening when lying down,¹² while frequently being located in the frontal or peri-orbital regions. Patients were advised to resume their daily activities gradually, starting with light tasks and progressively increasing the intensity and duration as tolerated, to ensure a safe and steady recovery. Patients underwent further clinical and radiological follow-up evaluations at 6 months, 1 year, and 24 months.

CSFVF diagnosis

A detailed description of our institutional protocol for localizing the CSFVF has been published.⁷ Basically, CSFVFs were detected on DSM performed under local anesthesia, exploring both sides of the spine on separate days, on an Azurion 7 B20/15 fluoroscopy unit (Philips Healthcare, Best, The Netherlands). A dedicated spinal myelography setup, with 3–4 frames/s and a tiltable table, was utilized to position the patient’s hips above the shoulders. A 20-gauge spinal needle was used to puncture the lumbar thecal sac. The table was tilted between 4° and 7° in the Trendelenburg position, and up to 20 mL of Iopamiron 300 was injected. A small contrast injection of 0.5 mL was administered to verify the needle’s intrathecal position. Subsequently, 7 mL of Iopamiron 300 was injected over 5 s, with the imaging system centered on the cervicothoracic segments. This was followed by a second injection of 5–6 mL of Iopamiron 300 targeting the thoracolumbar spine.

In cases where lower lumbar or sacral CSFVF was suspected (eg, the presence of perineural cysts in the sacrum), an additional 4 mL of contrast was injected while tilting the table in the anti-Trendelenburg position. During imaging DSM, patients performed resisted inspiration to maximize the visualization of the fistula.² During imaging with DSM, patients performed resisted inspiration to enhance the visualization of the fistula. This was achieved by asking the patient to take a deep breath through a 10 mL syringe during the procedure. The syringe, without the plunger, was held in one hand—using the right hand if lying right-side down or the left hand if lying left-side down—and the barrel flange was placed in the mouth. The patient was asked to breathe in through the syringe, starting the inspiration when the contrast had filled at least three-quarters of the spine of interest. Resisted inspiration was performed twice: once during the DSM of the upper cervicothoracic spine, and once during the DSM of the lower thoracolumbar spine. Cone beam CT (CBCT) was performed after the final injection to further aid in CSFVF detection. The patient was instructed to hold their breath during the CBCT to minimize motion artifacts. Immediately after, patients were transferred to the CT receiving a full-spine post-DSM lateral decubitus dual-energy CT. DSM analysis and image interpretation were conducted by a single neuroradiologist with expertise in the diagnosis and treatment of SIH. Among a total of 63 CSFVFs, 36 (57.1%) were detected using both DSM and dual-energy CT post-myelography, 22 (34.9%) were identified with both DSM and CBCT, and 18 (28.6%) were diagnosed with both CT and CBCT. Additionally, 17 (27.0%) CSFVFs were visible across all three diagnostic methods. Exclusively, 13 (20.6%) CSFVFs were diagnosed using DSM, seven (11.1%) were identified solely through post-myelography dual-energy CT, and none were diagnosed exclusively with CBCT.

Endovascular CSFVF treatment

The indication for endovascular treatment was confirmed during a multidisciplinary meeting involving a neuroradiologist, a headache neurologist, and a neurosurgeon, all with expertise in SIH. The treatment was performed under general anesthesia. Intravenous heparin was administered following catheterization of the venous system, achieving measured activated clotting times of 200–250 s. Access was obtained through the right or left common femoral vein. A 7 French (7F) Rist catheter (Medtronic) was advanced into the azygous vein, ascending the lumbar vein, or vertebral veins, based on the location of the fistula, over a 5F Merit vertebral catheter (Impress). Microcatheterization of the paraspinal vein was obtained either with a Headway Duo microcatheter or a Scepter XC microcatheter over a 0.14 inch microguidewire. Onyx 18 was used to embolize the targeted veins. Coils were used in some cases to occlude potentially dangerous communications with the azygos vein, preventing Onyx migration into the vein. Additionally, coils were used to occlude the posterior intercostal vein, limiting the unnecessary spread of Onyx within this vein.¹³

Statistical analysis

Categorical data were described by their frequency of occurrence, while quantitative data were summarized using their mean±SD. A χ² test was used to evaluate qualitative factors, and a two-tailed t-test was applied to assess quantitative factors. Univariate analysis was performed to identify potential factors associated with improvement at 24 hours and at 3-month follow-up. The variables analyzed included mean age, sex, body mass index (BMI), presence of multiple CSFVFs, number of CSFVFs, symptom duration, prior history of headache, SIH score before treatment, and improvement of SIH score at 24 hours (defined as the normalization of at least one of the six parameters included in the SIH score) (online supplemental table 3).

The interobserver agreement for the SIH score, SSS diameter, and TVS diameter was evaluated using the κ test to determine the consistency between observers. Measurements were taken at three time points: before treatment, at 24 hours post-treatment, and at 3 months post-treatment. The κ statistic was used to quantify the level of agreement, with values interpreted as follows: <0.20 (poor agreement), 0.21–0.40 (fair agreement), 0.41–0.60 (moderate agreement), 0.61–0.80 (substantial agreement), and >0.81 (almost perfect agreement). Statistical significance was set at P<0.05. All analyses (descriptive and inferential) were conducted using Statistical Package for the Social Sciences (IBM SPSS) version 30.

Results

Baseline population characteristics

Since November 2022, 60 consecutive patients with confirmed CSFVFs underwent endovascular treatment via transvenous Onyx injection. Of these, 40 patients underwent clinical evaluation and brain MRI at 24 hours post-treatment, as well as a mid-term follow-up MRI at 3 months; these patients were included in the analysis. Further clinical follow-up was conducted over an average of 12±4.4 months, with an average MRI follow-up of 8±4.1 months.

The mean patient age was 61±13 years, with 26 females (65%). The mean BMI was 24±4 kg/m², and the median duration of symptoms was 10 months (IQR 2–36). A prior history of headache was reported in 17 patients (42.5%), including six with episodic headaches (15%) and 11 with chronic headaches (27.5%) (online supplemental table 1). Fatigue was the most prevalent symptom (87.5%), followed by activity-related headache (72%), orthostatic headache (70%), dizziness (65%), Valsalva-related headache (62%), tinnitus (45%), and non-orthostatic headache (10%) (table 1).

Table 1. Frequency of symptoms before treatment and improvement rates at 24 hours and 3 months following CSFVF embolization.

Variables	Before treatment	Patients with resolution/improvement at 24 hours	Patients with resolution/improvement at 3 months	P value (before vs 24 hours)	P value (before vs 3 months)	P value (24 hours vs 3 months)
Fatigue	35 (87.5%)	10 (28.5%)	32 (91%)	0.009	<0.01	<0.01
Tinnitus	18 (45%)	2 (11%)	11 (61%)	0.6	<0.01	0.02
Dizziness	26 (65%)	18 (30%)	20 (77%)	0.08	<0.01	0.51
Orthostatic headache	28 (70%)	19 (68%)	25 (89%)	<0.01	<0.01	0.06
Non-orthostatic headache	4 (10%)	0 (0%)	2 (50%)	1	0.43	0.43
Activity-related headache	29 (72%)	24 (83%)	24 (83%)	<0.01	<0.01	1
Valsalva-related headache	25 (62%)	20 (80%)	21 (84%)	<0.01	0.01	0.72

Variables	Improvement	Total resolution	No change
Number of patients at 24 hours	31 (77.5%)	20 (50%)	9 (22.5%)
Number of patients at 3 months	38 (95%)	33 (82.5%)	2 (5%)

CSFVF, cerebrospinal fluid venous fistula.

A non-targeted epidural blood patch (EBP) had been administered to 13 patients (32%) before definitive treatment, with an average of 1.5 EBPs per patient. Transient clinical improvement following EBP was observed in five patients (38%), with a mean duration of symptom relief of 1 month.

CSFVF characteristics

A total of 63 fistulas were identified among the 40 patients. Most lesions were right-sided (23/40, 57.5%). Multiple CSFVFs were detected in 17 patients (42.5%), including five patients (12.5%) with bilateral lesions.

Most of the fistulas were located between T7 and T12 (46/62, 74%). Cervical spine involvement was observed in four cases (6%), and another four cases (6%) involved the lumbar spine, including one affecting the right S2–S3 region.

Early and mid-term imaging findings after treatment

The median SIH score before treatment was 6 (IQR 4–8), decreasing significantly to 3.5 (IQR 2–5, P=0.01) at 24 hours, and further to 2 (IQR 0–4, P=0.004) at 3 months after transvenous CSFVF embolization (table 2). The median change of the SIH score at 24 hours and 3 months was 3 (IQR 0–4), and 4 (IQR 3–5), respectively. There was no further change of the global SIH score after 3 months. Focusing on the individual MRI brain signs, pachymeningeal enhancement was the most common one at diagnosis, observed in 87.5% of the cases, followed by venous sinus engorgement (75%), and effacement of the SSS (65%), the PPC (62.5%), and the MPC (60%). These abnormalities showed significant regression at 24 hours post-embolization: pachymeningeal enhancement regressed in over 50% of patients, venous sinus engorgement in approximately 40%, and the other MRI findings in about 30% (figures1 2).

Table 2. MRI signs before treatment, at 24 hours, and at 3 months post-CSFVF embolization.

Variables	Before treatment	24 hours after treatment	3 months after treatment	P value (before vs 24 hours)	P value (before vs 3 months)	P value (24 hours vs 3 months)
Median (IQR) and mean±SD SIH score	6 (4–8) 6.1±2	3.5 (2–5) 3.5±2.2	2 (0–4) 2±2	0.01	0.004	0.01
Median (IQR) and mean±SD change of SIH score		3 (0–4) 2.5±2	4 (3–5) 4±2.1
Mean±SD and median (IQR) SSS diameter	7±1 6.9 (6–7.3)	5.9±0.9 6 (5–6.7)	5.6±1 5.5 (5–6.3)	<0.01	0.003	0.06
Mean±SD and median (IQR) TVS diameter	7.7±1 7.6 (7–8.2)	6.9±1 7 (6–7.5)	6.5±1 6.6 (5.7–7)	<0.01	<0.01	0.04
Mean±SD and median (IQR) SSC diameter	3.4±2 3.5 (1.4–5)	4.4±1.9 4.5 (2–6)	4.8±1 4.9 (3–6.5)	<0.01	<0.01	0.02
Mean±SD and median (IQR) pre-pontine cistern diameter	4.4±1.6 4.5 (4–5.5)	5.3±1.5 5.3 (4.6–5.9)	5.3±1.3 5.5 (4.5–6)	<0.01	<0.01	0.42
Mean±SD and median (IQR) mamillopontine cistern diameter	5.6±1.4 6 (4–7)	6.5±1.4 6.9 (5.5–7.7)	6.7±1.1 7 (5.9–7.6	0.02	<0.01	0.61

Variables	Before treatment	Number of patients with regression of MRI signs at 24 hours	Number of patients with regression of MRI signs at 3 months	P value (before vs 24 hours)	P value (before vs 3 months)	P value (24 hours vs 3 months)
Pachymeningeal enhancement	35 (87.5%)	19 (54%)	28 (80%)	<0.01	<0.01	0.03
Venous sinus engorgement	30 (75%)	12 (40%)	22 (73%)	0.006	0.001	0.06
Effacement of the SSC	26 (65%)	8 (31%)	15 (57%)	0.08	0.01	0.48
Pre-pontine cistern effacement	25 (62.5%)	8 (32%)	14 (56%)	0.07	0.01	0.43
Mamillopontine distance ≤6.5 mm	24 (60%)	9 (37.5%)	12 (50%)	0.04	0.01	0.51
Subdural fluid collections	12 (30%)	6 (50%)	0 (100%)	0.1	<0.01	0.02

CSFVF, cerebrospinal fluid venous fistula; SIH, spontaneous intracranial hypotension; SSC, suprasellar cistern; SSS, superior sagittal sinus; TVS, transverse venous sinus.

Figure 1. (A–D) A patient in their 70s presented with a 12-month history of orthostatic headache. (A) Axial T1-weighted brain MRI with gadolinium enhancement shows bilateral pachymeningeal enhancement (yellow arrows), consistent with spontaneous intracranial hypotension. (B) A right-sided CSFVF at T5-T6 and T6-T7 was treated using transvenous Onyx embolization. (C) A 24-hour post-treatment brain MRI demonstrates regression of dural enhancement. (D) Follow-up at 3 months confirms stability of the dural enhancement regression. (E–H) A patient in their 60s with orthostatic headache and Valsalva-induced headache due to spontaneous intracranial hypotension. (E) Brain MRI reveals bilateral pachymeningeal enhancement. (F) Right T12-L1 and left T11-T12 CSFVFs were treated with transvenous Onyx embolization. (G) Brain MRI at 24 hours post-treatment shows complete resolution of pachymeningeal enhancement (yellow arrows) and an increased diameter of the lateral ventricles compared with pre-treatment images. (H) A 3-month follow-up MRI demonstrates stable regression of the dural enhancement and a slight further increase in the diameter of the lateral ventricles. CSFVF, cerebrospinal fluid venous fistula.

Figure 2. A patient in their 40s presented with orthostatic headache and vertigo, which began suddenly 1 month prior. (A) Sagittal T1-weighted gadolinium-enhanced brain MRI demonstrates a reduction in the suprasellar cistern (red arrow), pituitary enlargement, and venous sinus engorgement (orange arrow, indicating the straight sinus). (B) A right T9-T10 CSFVF was treated with transvenous Onyx embolization. (C) At 24 hours post-treatment, brain MRI shows normalization of the venous engorgement (orange arrow) and an increase in the diameter of the suprasellar cistern (red arrow). (D) At the 3-month follow-up MRI, there is a further increase in the diameter of the suprasellar cistern, an increase in the mamillopontine distance, and stable regression of the venous sinus engorgement (orange arrow). CSFVF, cerebrospinal fluid venous fistula.

Further improvement was observed at 3 months. Pachymeningeal enhancement and venous sinus distension resolved in approximately 70–80% of cases. Subdural fluid collections, present in 30% of patients at baseline, regressed in 50% of cases within 24 hours, though this change did not reach statistical significance. By 3 months, however, complete resolution of subdural fluid collections was achieved in all patients. When comparing MRI findings at 24 hours and 3 months, pachymeningeal enhancement and subdural fluid collections demonstrated significant further regression at mid-term follow-up.

The diameters of the cisterns normalized in about 50% of patients over follow-up compared with the pre-treatment status. The mean diameter of the SSS decreased significantly when compared with pre-treatment, from 7±1 mm to 6±1 mm (P<0.01) at 24 hours, and from 7±1 mm to 5.6±1 mm (P=0.003) at 3 months. Similarly, the mean diameter of the TVS was reduced when compared with pre-treatment, from 7.7±1 mm to 6.9±1 mm (P<0.01) at 24 hours, and from 7.7±1 mm to 6.5±1 mm (P<0.01) at 3 months. The mean SSC diameter increased significantly when compared with pre-treatment, from 3.4±2 mm to 4.4±1 mm (P<0.01) at 24 hours, and further to 4.7±1 mm (P<0.01) at 3 months. Similarly, the mean PPC diameter increased significantly compared with pre-treatment, from 4.4±1.6 mm to 5.3±1.6 mm (P<0.01) at 24 hours, and maintained this value at 3 months. The MPC distance also showed a significant increase when compared with pre-treatment, from 5.6±1.4 mm to 6.5±1.4 mm (P=0.02) at 24 hours, and further to 6.7±1 mm (P<0.01) at 3 months.

Inter-rater reliability

The inter-rater reliability was assessed using Cohen’s κ coefficient across multiple parameters and time points. For the SIH score, the κ values demonstrated substantial agreement before treatment (online supplemental figure 3), as well as at 24 hours (online supplemental figure 4) and 3 months after treatment (online supplemental figure 5). Similarly, the SSS diameter showed consistent agreement pre-treatment (online supplemental figure 6), 24 hours post-treatment (online supplemental figure 7), and 3 months post-treatment (online supplemental figure 8). Lastly, the dominant TVS diameter exhibited moderate to substantial agreement across pre-treatment (online supplemental figure 9), 24 hours (online supplemental figure 10), and 3 months post-treatment (online supplemental figure 11).

Early and mid-term clinical outcomes

Patients were evaluated at 24 hours and 3 months post-treatment (table 1). At the 24 hour follow-up, 31 patients (77.5%) reported symptom improvement, with 20 (50%) describing complete resolution. Nine patients reported no clinical improvement (22.5%). Significantly improving symptoms at 24 hours included fatigue, dizziness, orthostatic headache, activity-induced headache, and Valsalva-related headache. However, tinnitus and non-orthostatic headache did not exhibit notable changes 24 hours post-treatment. At 3 months follow-up, 95% of the patients (38/40) showed significant clinical recovery, of which 86.8% (33/38) achieved complete clinical recovery. Two patients (5%) reported no clinical improvement (both were female and had non-orthostatic headaches and tinnitus for more than 3 years). At mid-term follow-up only non-orthostatic headache did not change significantly when compared with the pre-treatment status.

Four patients (10%) required a second embolization due to recurrent symptoms. Among these, three developed a new CSFVF, with the median time to symptom recurrence and diagnosis of the new fistula being 13 months (IQR 13.5–21.5). One patient required an additional Onyx injection after 6 months to treat the incomplete occlusion of a previously treated C7–T1 lesion. This patient experienced partial improvement in orthostatic headache and dizziness, which recurred after 6 months when a new DSM revealed an incompletely occluded C7–T1 lesion.

Predictors of clinical improvement at 24 hours

A significant association was observed between the improvement of SIH at 24 hours and clinical improvement within the same timeframe, with 81% of patients demonstrating clinical improvement when SIH improved, compared with only 19% without SIH improvement (P=0.002) (online supplemental table 3). However, this association was not maintained at the 3-month follow-up. At that point, 63% of patients with SIH improvement at 24 hours exhibited sustained clinical improvement, compared with 50% of those without SIH improvement; however, this difference was not statistically significant (P=0.99). Furthermore, none of the variables studied at the 3-month follow-up showed a significant association with the investigated outcome (online supplemental table 3).

Technical outcomes and procedure-related complications

Embolization was technically successful in all cases (online supplemental table 2). The median procedure duration was 45 min (IQR 30–70). Nine patients (22.5%) experienced transient local pain at the injection site, which resolved within an average of 3 days. In two cases, the microcatheter perforated the epidural plexus without causing clinical complications or affecting the procedure’s outcome. Additionally, asymptomatic Onyx migration into the azygos system was observed in two patients. We did not observe any permanent morbidity or mortality.

Post-treatment rebound headache

Thirteen patients (32.5%) experienced PRH (online supplemental table 2), characterized by worsening headaches when lying down, primarily localized to frontal and periorbital regions. PRH typically occurred within the first 24–48 hours after treatment, with a mean duration of 7±3.5 days. Nine patients required medical management with oral acetazolamide, initiated at a starting dose of 750–1000 mg/day, and continued for approximately 1 week until the resolution of PRH.

Discussion

In this study we were especially interested in early MRI and clinical outcomes at 24 hours after transvenous CSFVF embolization. To our knowledge, this is the first study to evaluate such rapid changes in relation to mid-term improvement at 3 months. Forty consecutive patients with more than 60 CSFVFs treated with Onyx embolization underwent MRI evaluations at 24 hours and 3 months post-treatment. This highlighted several key findings. First, transvenous embolization proved to be rapidly effective, demonstrating significant improvement in brain anomalies within the first 24 hours. Clinically, over three-quarters of the patients reported symptom improvement at 24 hours, with half of them experiencing complete resolution, especially of symptoms such as orthostatic headache, or activity-induced or Valsalva-induced headache. In contrast, tinnitus, dizziness, fatigue, and non-orthostatic headache showed slower and less pronounced rates of improvement. Finally, the regression on any of the MRI brain signs at 24 hours was associated with early clinical improvement.

Early radiological and clinical outcomes

The main finding of our study was the significant reduction in the SIH score within 24 hours post-embolization, with its median score decreasing by approximately half, from 6 to 3.5. Furthermore, the SIH score continued to decrease over follow-up, reaching a median of 2 at 3 months post-embolization, despite most of the improvement occurring within the first 24 hours. These results are important because they demonstrate the effectiveness of Onyx transvenous embolization in successfully closing CSFVFs, regardless of whether the fistulas are single, multiple, or bilateral. Our 3-month MRI findings align with previously published studies.13,15 For example, in the largest series to date on CSFVF, Brinjikji et al⁸ reported over 80% improvement in SIH scores at mid- to late-term follow-up compared with baseline, with a mean reduction of 5 points.

To our knowledge, no other series has investigated early changes after CSFVF treatment. There is only one recent case report of Saionz et al⁹ documenting rapid clinical and MRI improvement within 48 hours following CSFVF embolization. Accordingly, we confronted our data with early findings after type I and II leaks. Indeed, Dobrocky et al¹⁶ reported clinical and MRI outcomes 4 days post-surgical repair of 19 SIH patients. Like us, the mean preoperative SIH score decreased significantly, from 6.9 to 2.9, within this short time frame after surgical closure of the dural breach. Similarly, 90% of patients exhibited early symptom improvement within a few days of surgery, highlighting the rapid clinical benefits of the targeted reparation of SIH, independently of the type of leak. This may not be surprising when considering that the rate of CSF production is approximately 18–25 mL per hour, amounting to 430–530 mL per day, and that elevated CSF production may be observed in patients with SIH,⁴ counterbalancing quickly the lack of fluid once the leak has been sealed.

Our analysis also revealed that certain MRI markers included in the SIH score exhibit more pronounced changes within 24 hours than others. Pachymeningeal enhancement resolved in over 50% of patients undergoing transvenous CSFVF embolization. Similarly, venous sinus engorgement and subdural fluid collections regressed in approximately 40–50% of patients. In contrast, the diameters of the SSC, PPC, and MMP cisterns normalized in about one-third of patients by the first day post-embolization.

Overall, clinical improvement rates were approximately 78% at 24 hours post-treatment. However, the rate of improvement varied depending on the type of symptom. Headaches transiently exacerbated or triggered by Valsalva maneuvers and/or orthostatic changes, which affected 60–70% of patients, showed significant change the day after treatment, with early improvement rates between 70% and 80%. These rates further increased to 80–90% at the 3-month follow-up. However, approximately 30% of patients experienced PRHs, characterized by the disappearance of orthostatic and/or Valsalva-related components. These cases were classified as regression of SIH-related headaches during the early follow-up period.

In contrast, non-orthostatic headaches—characterized by the lack of relief with recumbency—affected 10% of the cohort. None of these patients reported improvement during the early post-embolization period, and only 50% experienced symptom relief by the 3-month follow-up. As the disease progresses, the characteristic orthostatic nature of the headaches seems to evolve into a more generalized, chronic non-orthostatic headache pattern. This previously orthostatic headache has become independent of pressure fluctuations and is thus likely less responsive to leak repair, reflecting a potential shift in the underlying pathophysiology.^{5 17}

We acknowledge the possibility that rebound intracranial hypertension (RIH) or underlying idiopathic intracranial hypertension (IIH) may contribute to the incomplete headache response observed in some patients. Although RIH has been proposed as evidence of elevated intracranial pressure being the primary cause of CSFVF, we find this explanation less convincing for our cohort. Notably, a significant proportion of our patients do not display the typical demographics or comorbidities associated with the IIH population, such as obesity or female sex.

In our cohort, the median BMI was 23.8, 65% of the population were female, and the median age was 61 years. By contrast, in typical IIH populations, 80–90% of cases are female, the median age of diagnosis is close to 30 years, and the mean BMI is approximately 35 or higher.¹⁸ Furthermore, while incomplete headache resolution may suggest residual intracranial hypertension, alternative mechanisms should be considered. These include altered CSF dynamics or the presence of chronic migraine.

Less than 30% of patients experienced improvement in tinnitus, dizziness, and fatigue during the early post-embolization period, with up to 40% still reporting tinnitus at follow-up. Similarly, the persistence of tinnitus and hearing disturbances was noted by the Mayo Clinic group, who reported these symptoms in up to 36% of patients undergoing transvenous Onyx embolization for CSFVF.⁸ Tinnitus may arise from the transmission of abnormal CSF pressure to the cochlear perilymph or from brain sagging, which can stretch and irritate the vestibulocochlear nerve, leading to auditory symptoms such as tinnitus that may persist despite resolution of the cause.¹⁹

Factors associated with early and mid-term clinical improvement

Since the available literature is limited, gathering evidence on predictive factors for clinical improvement after CSFVF embolization is crucial. Our analysis revealed that neither patient-related factors, CSFVF characteristics, nor the severity of brain MRI findings correlated with early clinical improvement. The only significant predictor of early improvement was any change in SIH score within 24 hours post-treatment. Approximately 80% of patients with clinical improvement at 24 hours showed a reduction in their SIH score, while nearly 80% of those without improvement exhibited no early score change (online supplemental table 3). Patients often reported significant improvement in positional or Valsalva-related headaches the day after embolization, with 70–80% experiencing symptom regression early on. However, at the 3-month mark, none of the analyzed factors, including the SIH score, was associated with clinical outcomes. Interestingly, this lack of correlation between SIH score and clinical severity has been previously noted at the pre-treatment stage.²⁰ The authors suggested that poor clinical predictive status of brain imaging SIH signs might be time related. That is, imaging findings tend to become less pronounced over time, in line with the previously described shift in underlying pathophysiology,^{5 16} that potentially weakens the correlation between imaging and clinical symptomatology.

Importantly, 65% of patients with improvement at mid-term follow-up also exhibited rapid SIH score changes within 24 hours (online supplemental table 3), suggesting early improvement as a potential predictor of long-term outcomes. However, our study lacks the statistical power to confirm the impact of early MRI changes on final clinical outcomes. We could only confirm the association between both at the same early time-frame. Other studies have identified a previous history of headaches and female sex as factors reducing the likelihood of a complete response following CSFVF embolization, but we could not confirm this.⁸ Additionally, delayed diagnosis and treatment have been associated with a lower chance of achieving complete clinical remission for both type I–II and type III leaks.^{8 21}

Limitations

This study has limitations. First, some analyzed factors may not reach the statistical significance due to the relatively small sample size. Accordingly, we will continue to collect data on early and mid-term clinical and MRI changes in patients treated with transvenous CSFVF embolization. Second, the analysis was conducted retrospectively, despite the database being maintained prospectively.

Conclusions

The significant decrease in SIH score, approximately twofold within 24 hours following transvenous embolization CSFVF, highlights the high efficacy of this treatment in sealing the leak. This translates to an approximate 80% rate of clinical improvement observed the day after the procedure, primarily driven by the resolution of orthostatic and Valsalva-related headaches. The SIH score continues to improve, showing a threefold reduction at the 3-month follow-up and stabilizing thereafter. Similarly, patients exhibit progressive recovery, with >80% achieving complete resolution of symptoms. However, tinnitus and non-orthostatic headaches persist in approximately 50% of cases.

Data availability statement

Data are available upon reasonable request.