Spontaneous cerebrospinal fluid (CSF) leak is a rare clinical entity that may result in disabling headaches. It occurs as a result of dural defects, and the initial symptoms resemble those of postdural puncture headache. However, the positional headache can later evolve into a persistent chronic daily headache. The diagnosis of spontaneous CSF leak can be very challenging, but increasing awareness and improved diagnostic techniques are yielding ever more cases. When conservative management fails, the pain management clinician is called upon to administer an epidural blood patch. The success of this technique is dependent upon accurate diagnosis of the site of leakage and targeted epidural administration of the blood patch to this area. In this report, we describe four consecutive cases that were referred to our pain management department over an 18-month period and were successfully treated with site-directed epidural blood patches.
Low cerebrospinal fluid (CSF) pressure headache is frequently encountered in pain clinics following dural puncture. Spontaneous CSF leak headache, however, is a rare entity, and its diagnosis and management can pose serious clinical challenges. We report on the management of four consecutive cases that were referred to our pain management department over an 18-month period and review of the literature on the topic.
A 46-year-old construction worker developed an occipital headache and left-sided neck pain in late February 2001 while walking with his wife. Over the subsequent weeks, the headache continued to recur when the patient was sitting or standing and was relieved by recumbency. Initial workup by his family physician with CT scan and MRI of the brain and cervical spine was unremarkable. A neurologist evaluated him, and a diagnosis of low CSF pressure headache was made. A CSF opening pressure of 80 mm H2O was obtained by lumbar spinal tap; however, the patient's symptoms were exacerbated, and he developed nausea, vomiting, and dehydration. He was admitted for intravenous hydration and underwent a lumbar epidural blood patch. He was discharged home after adequate rehydration and resolution of the nausea but had a persistent orthostatic headache. A repeat MRI, with and without contrast, revealed increased meningeal enhancement over the cerebral convexities and involving the falx and tentorium as well as prominent meningeal enhancement throughout the cervical region. MRI of the thoracic spine revealed mild T8-T9 disc bulging but no contrast enhancement of the meninges. An 111Indium radioisotope cisternogram was performed and did not show any extravasation of activity from the thecal sac to suggest CSF leak. Of note, the CSF flow was extremely slow to allow for detection of the CSF opening pressure during myelography. Subsequently, the patient was referred to the Cleveland Clinic Foundation, where a CT myelogram with delayed imaging of the entire spine was obtained. It revealed a curvilinear collection, 1 cm long and 1 mm thick, in the dorsal epidural space at T6-T7 and a similar one at T7-T8 (Figure 1). A consultation with the Pain Management Department was initiated for a possible epidural blood patch procedure. A T7-T8 thoracic epidural blood patch with 20 cc of autologous blood was performed using a paramedian approach with fluoroscopic guidance. An MRI of the thoracic spine, performed immediately following the blood patch, showed a dorsal epidural soft tissue extending from the T3 to the T8-T9 level, likely representing the injected blood (Figure 2). Two days after the procedure, the patient's headache was completely resolved, and he continued to do well ever since and had returned to work as of June 4, 2001.
A 55-year-old woman with no significant head-ache history developed a sudden severe occipital headache doing housework in June 2002. The next morning she awoke without a headache, but when she arose from bed, she experienced severe pain in the head, and from that point she continued to have a positional exacerbation of her constant headache. No focal motor, sensory, or coordination disturbances were noted on neurologic exam. She tried multiple medication regimens without relief. A greater occipital nerve block provided only limited relief. On MRI, there was diffuse enhancement of the thickened dura, findings compatible with CSF leak (Figure 3). A small bilateral subdural fluid collection, which showed increased T2 and isointense T1 signal, was noted and was thought to reflect recent hemorrhage or a fluid collection with high protein content. A CT myelogram revealed CSF leak coming from the area around the root sleeve at C7-T1 where the contrast extravasates out through the nerve root into the epidural space. In addition, the patient did have an epidural lipoma running from T2-T8. A cervical epidural blood patch was performed in August 2002, using fluoroscopic guidance and 20 cc of autologous blood at C7-T1. After the procedure, the patient experienced 50% pain, but a low-grade headache at the top of her head and neck persisted. She failed treatment with IV caffeine, ketorolac, and magnesium infusion and was referred for a second epidural blood patch. Her second blood patch was performed in October 2002 under fluoroscopy at C7-T1 using 20 cc of autologous blood. The patient left the clinic with no pain, and her headache has resolved since.
A 36-year-old salesman developed a headache spontaneously one morning in September 2001 and, as the day progressed, his pain was worsening, with bifrontal and bioccipital nuchal distribution. Over the next few days, the patient noticed a positional component to the headache. Associated symptoms included nausea and lightheadedness, but he had no focal motor sensory disturbances. Previous testing has included several computed axial tomography (CAT) scans of the brain. MRI, magnetic resonance angiography (MRA), and magnetic resonance venography (MRV) of the brain were reported as normal. Previous abortive medication attempted included NSAIDS, mild opioids, and inhaled triamcinolone, all without relief. His neurologic exam was unremarkable. Diagnostic nasal endoscopy along with radionuclide CSF tracer revealed no evidence of CSF rhinorrhea. The CSF pressure prior to injection of the tracer was noted as “low” with scant CSF return. A radioisotope cisternography study was also read as normal, however, there was some delay in the tracer reaching the convexities. After the patient failed IV caffeine, he was referred to the Pain Management Department for an epidural blood patch because of the positional nature of the headache and the low CSF pressure. An epidural blood patch was performed at L1-L2, 1 month after the start of his symptoms, with 20 mL of autologous blood, and resulted in a 50% pain relief for 2 days, then his headaches returned. The patient then underwent a CT myelogram of the entire spine to localize potential CSF leak. The CT myelogram was grossly abnormal with evidence of extensive subdural fluid collection in the midthoracic spine from T3 to T9 and a small amount of epidural contrast, suggesting this represented the primary site of CSF leak. In addition, there was a small amount of epidural contrast that related to a second site of leak, which was around T11. Furthermore, there was evidence of an eccentric disc extrusion and a free disc fragment at T4-T5 with moderate cord compression. The patient was referred for a second blood patch that was performed with C-arm guidance at T5-T6 using 10 cc of autologous blood. The patient was discharged with no headache, and pain relief persisted in follow-up visits.
A 31-year-old female nurse developed gradual onset of bifrontal headache associated with neck pain in November 2001. The patient did not have a history of migraines and the headache and was not associated with phonophobia, photophobia, or aura. Initially, her headache was achy, global, and nonthrobbing with a positional component. She was treated with a number of medications, but only topiramate and naprosyn relieved the headache partially. An evaluation by the ENT revealed no nasal or sinus abnormality. The patient underwent an MRI, MRA, and MRV, which did not define an etiology of these headaches. In particular, there was no evidence of leptomeningeal enhancement. CT myelography showed a subtle irregularity of the thecal sac immediately dorsal to the nerve root sleeves at the C4-C5 and C5-C6 levels to either side of the midline, which may represent mild extravasation of contrast. No distinct epidural fluid collection was noted in this region. However, CSF pressure was 16 cm H2O. Following myelography, she had a postdural puncture headache, which manifested as exacerbation of the current headache. This exacerbation was successfully treated with a lumbar epidural blood patch, but her original headache persisted. Given the positional nature of the headaches, the patient was referred for an epidural blood patch at C5-C6 in July 2002. This was performed under fluoroscopy, and 7 cc of autologous blood was injected at that level. There was no immediate relief of the headache, but over the next 3–5 weeks, the patient noticed at least an 80% decrease in her headaches, which has continued to improve since.
Orthostatic headache is a consequence of CSF leakage and intracranial hypotension. Intracranial CSF pressure is dependent on CSF volume and posture. In the erect posture, the intracranial CSF pressure is negative compared to the lumbar pressure of 60–180 mm H2O. A decrease of CSF volume by 10% is sufficient to induce orthostatic headache . The orthostatic headache is a consequence of the low CSF pressure producing downward displacement of the brain causing traction on supportive pain-sensitive structures. The most common cause of postural headache is dural puncture following spinal tap or surgery. However, increasing cases of spontaneous intracranial hypotension are being recognized as a result of higher awareness and improved detection practices. The CSF leak, in these cases, is spontaneous and occurs most commonly in the thoracic spine or the cervicothoracic junction [2–4]. A connective tissue disorder or abnormality was identified in a substantial proportion of patients with CSF leaks . The cardinal manifestation of intracranial hypotension is headache in the upright posture that is relieved by assuming recumbency. Additional signs and symptoms may include neck and shoulder pain or stiffness, nausea, vomiting, photophobia, sixth cranial nerve paresis resulting in horizontal diplopia, tinnitus, and, rarely, radicular arm symptoms [4,6–8]. Diagnostic findings include a low opening CSF pressure on spinal tap, CSF pleiocytosis, and increased protein concentration as well as characteristic findings on radiological studies and radioisotope cisternograms. In the latter test, accumulation of radioactivity outside the subarachnoid space and failure of the tracer to reach the convexities of the brain provides an indication of a CSF leak [2,9–11]. In addition, rapid disappearance of the radioisotope from the subarachnoid space and early appearance in the urinary bladder suggest CSF leak and its early reuptake into the venous system. The false-negative rate for radioisotope cisternography, however, is high [7,12]. MRI findings include diffuse symmetric smooth dural thickening and pachymeningeal gadolinium enhancement, not only at the cranial level, but also involving the cervical spinal dura [11,13–17]. The detection of meningeal diverticula is common in these patients and does not always correlate with the site of CSF leak. MRI and CT myelography are the most reliable tests to determine the site of CSF leak [18–20]. The meningeal enhancement and subdural fluid collection result from venous hypervolemia consequent to CSF hypovolemia [21,22].
Though classical presentations of spontaneous intracranial hypotension may pose a modest diagnostic challenge, a broader syndrome is gaining recognition as varied manifestations of the syndrome are encountered. It is not unusual, after some period of time, for the orthostatic headache to change in nature and become a lingering, constant, and diffuse headache regardless of posture. In addition, normal CSF pressures have been recorded in patients with typical clinical and radiological features. Likewise, patients with identifiable CSF leak and typical orthostatic headache with low CSF pressure may not exhibit pachymeningeal enhancement. In a few patients, no orthostatic headache is perceived despite typical imaging features and low CSF pressures. For these reasons, Mokri suggested replacing the term intracranial hypotension with CSF hypovolemia, as low CSF volume is the unifying feature in all patients .
Spontaneous intracranial hypotension is often a self-limited condition and, in many patients, improves with conservative management. This includes bed rest, hydration, an abdominal binder, and caffeine. Invasive measures are considered when conservative management fails, and these comprise epidural blood patch as well as surgical repair [4,6]. A “directed” epidural blood patch, placed in close proximity to the level of suspected CSF leak, is effective in most patients. If symptoms recur, a repeat procedure may be performed. A recent report suggested epidural fibrin glue application should epidural blood patching fail . If reasonable attempts at targeted epidural procedures fail to control the symptoms, then surgical interventions may become necessary. These include various measures aimed at sealing the CSF leak with sutures, muscle packing, and blood and fibrin gluing [4,7].
The mechanism of headache relief following epidural blood patch has been investigated clinically and experimentally [24–27]. Not only does an epidural blood patch result in rapid onset of relief, the abatement of symptoms is long lasting in the majority of cases of post dural puncture headache . The rapid onset of pain relief following an epidural blood patch is believed to be due to an initial hydrostatic or “mass effect”[26–28]. Simple occlusion of the dural defect cannot account for the immediate relief of the symptoms because the rate of CSF production (0.35 ml/min) is too slow to correct the CSF hypovolemia in such a short time [26,29]. However, this effect is short lived. For the therapeutic response to be long lasting, sealing of the dural tear with a clot is required [26,27]. The coagulability of the blood is essential as EDTA-containing blood, saline, dextran-40, and hetastarch did not result in persistent CSF pressure elevation in a rat model of dural puncture . The leaking CSF may favor clot formation at the site of dural tear. Even though blood injected in the epidural can spread to multiple levels [26,30] as seen in case 1 in this study, the majority of the clot and mass effect appears to concentrate in the area around the injection site . This constitutes the basis for the recommendation that the blood patch be performed at one to two spinal segments below the level of dural tear in cases of post dural puncture headache . Our findings support such a recommendation for site-directed epidural blood patch in cases of spontaneous CSF hypovolemia with headache.