Epidural Blood Patching in Spontaneous Intracranial Hypotension—Do we Really Seal the Leak?

Medical records of consecutive SIH patients with a myelographically confirmed CSF leak, treated with a non-targeted EBP at our institute between January 2012 and November 2019 were reviewed. Exclusion criteria were a lack of a CSF leak on spinal imaging or if the patients have had previous spinal interventions like surgery.

Age, sex, symptoms (typical or atypical), number of EBPs performed, the volume injected into the epidural space, imaging studies, and surgical reports after a failed EBP were retrieved from the local clinical database and the picture archiving and communications system. Typical symptoms of SIH were defined as orthostatic headache and additional symptoms associated with SIH, such as neck pain and tinnitus. Atypical symptoms were defined as nonpostural headache patterns, absence of headache and dominance of other associated symptoms (e.g., nausea, diplopia, hearing loss).

All patients fulfilled the SIH criteria according to the International Classification of Headache Disorder (ICHD [12]) and demonstrated a spinal longitudinal epidural CSF collection (SLEC) on multimodal spine imaging including MRI, conventional dynamic myelography, or dynamic CT myelography (Fig. 1; [7]).

According to our clinical standard, a non-targeted EBP was performed when conservative measures failed to provide long-term relief. Repeated EBP had been discussed interdisciplinarily with the referring neurosurgeon or neurologist, typically when clinical symptoms improved but did not completely resolve after the first EBP.

The effect of the EBP and the extent of the postinterventional change in symptoms was evaluated from the clinical assessment reports of the attending physicians during a follow-up visit. Complete relief of symptoms was defined as resolution of all symptoms associated with SIH. Partial improvement was defined as persistent but attenuated symptoms. Patients whose symptoms completely resolved or who reported a partial improvement were referred to as responders. Non-responders were patients with unchanged symptoms or clinical worsening (Fig. 2). The clinical outcome was summarized as short-term outcome, correlating the clinical status of the patients within the 7 days following the EBP, with long-term outcome defined as the patients’ health status after the third postinterventional week.

Further steps after clinical or imaging follow-up were decided for each patient individually and included additional EBP, surgical intervention or further conservative treatment.

The anonymized data were edited in REDCap (Research Electronic Data Capture) and evaluated with descriptive statistics.

Brain MRI with intravenous gadolinium was acquired on a 1.5-Tesla or 3‑Tesla scanner. Spine MRI was acquired on a 1.5-Tesla scanner (Aera; Siemens, Erlangen, Germany) using a 20-element head/neck coil and a 32-element table spine coil. As previously described, the unenhanced spine MRI protocol included 3 sagittal T2-weighted spin-echo blocks (TE 110, TR 4350, time of acquisition: 03:33 min:sec, flip angle: 150°) and 4 sagittal, isotropic 3D heavily T2-weighted turbo spin-echo sequences with fat saturation (TE 180, TR 1400, time of acquisition: 04:33 min:sec, flip angle: 150°) including the entire neurocranium [13].

Conventional or dynamic CT myelography was performed during intrathecal injection of non-ionic contrast agent to localize the exact site of the CSF leak. This technique has been previously described [8, 14].

Briefly, for conventional dynamic myelography the mobile table was tilted into the Trendelenburg position, with the x‑ray detector following the leading edge of the intrathecal contrast. agent The goal was to identify the level where contrast started spilling from the intrathecal into the epidural compartment. The patient was then transferred to the CT imaging suite, and postmyelography CT (PMCT) was performed (Somatom Definition Edge, Siemens, Erlangen, Germany) to identify the underlying pathology (e.g., disc protrusion with an osseous spur) [13].

The non-targeted EBP procedure was performed on a monoplane high-resolution angiographic system (Artis zee multipurpose, Siemens, Erlangen, Germany) equipped with a flat panel detector (30 × 40 cm) with the patient in a sitting or lateral decubitus position. Under fluoroscopic guidance a 17-gauge epidural needle (Espocan®, B. Braun, Melsungen, Germany) was slowly advanced into the epidural space in the lumbar spine (L2/3, L3/4 or L4/5) with continuous pressure applied to the syringe plunger, known as the loss-of-resistance technique. The epidural needle was angled cranially to promote more rostral flow of the injected blood. The position of the needle tip in the epidural compartment was confirmed with a small amount of ionic contrast medium (Iopamiro 300, Iopamidol, Bracco, Cadempino, Switzerland). If the puncture was performed with the patient in the sitting position, the table was then tilted into the horizontal position and the sterile autologous blood was slowly injected. The added contrast agent demonstrated that the blood spread cranially along the spinal canal to the thoracic spine. The volume of injected blood was adapted to each individual and based on the patient feedback. The injection was paused when patients reported discomfort in the lower back or radiating pain and cramps in their buttocks and legs and resumed when these symptoms subsided. The injection was stopped, and the needle removed if slight discomfort persisted for an extended period. The patients were asked to remain in slight Trendelenburg position for 6 h to maximize distribution of the injected blood along the spine [15].

Owing to persisting symptoms in 28/51 (55%) patients after EBP therapy, spinal images (spine MRI and/or CT myelography) were acquired (Fig. 3). Mean time between EBP and imaging was 14 weeks (range 1–112 weeks). In 26/28 (93%) of these patients a SLEC+ was demonstrated after the last EBP. Two patients (7%) did not show any SLEC on post-EBP imaging. One had a ruptured nerve root cyst, diagnosed by dynamic myelography, and the MRI was acquired after the fourth EBP, and the second had multiple nerve root cysts and there was no evidence of SLEC after the second EBP. In addition, in 10/51 (20%) patients, EBP failed to provide clinical relief, and they were referred for surgery without additional imaging. In all these patients, a dural breach with an epidural CSF collection was present intraoperatively. In the remaining 13/51 (25%) patients, no postinterventional spine MRI and/or CT myelography was available and no surgery was performed.

In summary, a persistent spinal CSF leak was present after EBP in 36/51 patients (71%). Even in the best-case scenario accounting for missing data, if the leak had been successfully sealed in all patients with missing post-EBP spine MRI who had not undergone surgery, this would result in a maximum success rate of 29% (15/51 patients; 2 with a sealed leak verified on imaging (1 and 3 weeks after EBP) + 13 with no follow-up imaging).

The short-term outcome after the first postinterventional week was available for all enrolled patients (51/51, 100%). Overall, 45/51 (88%) patients responded to the first EBP in the short-term and only 12% (6/51) showed no improvement of the clinical condition (Table 2). None of the patients reported symptom worsening. If symptoms recurred after the first EBP, additional EBPs were performed according to the patient’s wishes and the decision of a multidisciplinary SIH board. Therefore, a total of 2, 3 or 4 EBPs were performed in 11 (22%), 1 (2%), and 1 (2%) patients, respectively (Table 1). Data for the long-term outcome were available for 47/51 (92%) patients with a mean follow-up period of 6.5 months (range, 1–50 months). Overall, 17/47 (36%) patients responded to the EBP in the long-term and 30/51 (59%) patients remained non-responders in the long-term. The main side effect reported was mostly low back pain for 1–4 days as well as headache and tinnitus in the first 24 h after treatment. No complications or long-term side effects were reported.

In total, 27/47 (57%) patients subsequently underwent surgical closure of the underlying CSF leak because of persisting symptoms after EBP. Of the 51 patients 4 (8%) remained non-responders but did not undergo surgery.

Even though untargeted EBP seems to be an effective symptomatic therapy in the acute stage, it rarely seals the underlying dural breach permanently. Thus, even if clinical improvement occurs in the early follow-up period, persisting CSF loss may perpetuate the disease and potentially result in serious long-term morbidity. This raises the question of whether untargeted EBP therapy is a causal or merely a symptomatic therapy. Should the permanent sealing of an underlying CSF leak in SIH patients be the main goal?

To maintain a stable intracranial volume, which is essential in a rigid structure such as the skull, CSF loss is compensated by a volume increase in other intracranial compartments [5]. Intracranial CSF depletion results in typical brain imaging findings such as dural enhancement, engorgement of venous sinuses, brain sagging, and subdural hematoma [1]. Consequently, CSF loss may result in serious conditions including subdural hematomas requiring decompressive craniotomy, non-convulsive status epilepticus, cognitive decline (frontotemporal brain sagging syndrome) or even coma [1, 24‐28]. Several studies have reported on patients with extensive superficial siderosis linked to a CSF leak of long duration [5, 29]. The proposed mechanism for this is the bleeding from small bridging veins, which have become stretched due to brain sagging. Thus, maintaining a CSF equilibrium appears to be important for preventing morbidity and preserving a good quality of life and may thus justify a more aggressive treatment approach.

A recent study reported that symptom duration before surgical closure of a CSF leak was the most important predictor of good outcome (improvement of orthostatic headache intensity). The authors concluded that treatment should be initiated within 12 weeks after symptom onset [30].

Our study cohort is comparable to those in previously published studies in terms of age, sex and CSF leak location (95% of leaks at the cervicothoracic or thoracic level) [1, 10, 16‐18]. Likewise, the EBP technique used during the study period is equivalent to that used in other centers. Usually, 10–20 ml of autologous blood was applied with the targeted or untargeted EBP method in the mentioned studies. However, the success rate of the untargeted EBP therapy in our study, defined as resolution of SLEC on follow-up imaging or intraoperatively and long-term clinical improvement, is lower than in most other studies.

There are many possible explanations for this difference. First, previous studies relied on clinical parameters only and not on imaging during follow-up. Proof of successful sealing of the dural breach was not required for evaluation of the EBP effect. Second, the inclusion criteria in our study were strict since only SIH patients with a proven CSF leak (on spinal imaging) were considered. Third, a change in headache phenotype or intensity after EBP therapy is considered as treatment success, which might not truly reflect the complex clinical presentation of SIH patients who often exhibit multiple symptoms. Fourth, the clinical presentation after EBP therapy does not always correlate with imaging findings [9]. In some of our patients, symptoms resolved after EBP even though the CSF leak was still present. Conversely, some patients reported persistent symptoms even though imaging showed no epidural CSF leak in spine MRI. This highlights the discrepancy between the clinical presentation and objective imaging proof of epidural CSF leak in SIH patients after EBP. The clinical presentation after treatment can pose diagnostic difficulties, may be subjective, and does not prove that the underlying spinal CSF leak is successfully sealed. Thus, evaluating the success of EBP therapy based on the clinical status without spinal imaging is insufficient.

In conclusion, our retrospective data analysis showed that in most patients with SIH due to CSF leak untargeted EBP alone is unlikely to improve patients clinically or seal spinal CSF leaks. This was confirmed on spinal imaging and/or intraoperatively and was also correlated with the clinical status of the patients. Thus, by performing EBP, which is often a symptomatic therapy with a low rate of successful sealing of the underlying dural breach, we might delay or even prevent adequate causal therapy and thus fail to avoid worsening of the disease. According to our data, patients with SIH due to radiologically proven CSF leaks should be actively investigated for the source of the leak and undergo subsequent early surgical intervention. In addition, the costs and the risks of the EBP therapy may not be in proportion to its limited effectiveness of the untargeted EBP therapy.

The major strength of this study is the carefully selected study cohort that included only SIH patients with a proven spinal CSF leak. Second, EBP procedures were carried out in a standardized fashion: image-guided and non-targeted, requiring confirmation of the correct position of the needle tip with contrast agent. There are some limitations inherent to the retrospective nature of the study, accounting for missing data and the small study size. Furthermore, no patients with a CVF were included, since EBP therapy has been previously demonstrated to provide poor results in patients with this condition [31].