Introduction

Spontaneous intracranial hypotension is an important and currently routinely diagnosed cause of new daily persistent headaches, particularly in young and middle-aged adults [1]. Children and the elderly, however, also can be afflicted [2, 3]. The most commonly diagnosed cause of spontaneous intracranial hypotension is a spontaneous spinal cerebrospinal fluid (CSF) leak, often in the setting of a generalized connective tissue disorder [1]. Alternative mechanisms include decreased CSF production, increased CSF absorption, pooling of CSF, and altered spinal dural elasticity. Importantly, there is no evidence to suggest any role of cranial base CSF leaks as the cause of spontaneous intracranial hypotension [4]. Although spontaneous intracranial hypotension is not a rare disorder and there has been a renewed interest in this cause of headaches since its MRI features were described in the early 1990s, an initial misdiagnosis remains common. The cardinal MRI features of spontaneous intracranial hypotension are sagging of the brain, enhancement of the pachymeninges, engorgement of venous structures, pituitary enlargement, and subdural fluid collections (mnemonic, SEEPS) (Table 1) [1]. The decrease of CSF volume in the subarachnoid spaces not only causes sagging of the brain but also manifests itself by a decreased diameter of the optic nerve sheath [5, 6]. Some patients with spontaneous intracranial hypotension have normal brain MRI findings, but the frequency of normal imaging is not really known, because the denominator is undetermined. In some series, as many as 20 % to 30 % of patients with SIH have a normal brain MRI [1].

Table 1 MRI characteristics of spontaneous intracranial hypotension
Full size table

Typically, a patient with spontaneous intracranial hypotension will present with an orthostatic headache. Associated symptoms are common, such as neck pain, nausea/emesis, hearing, or visual abnormalities, tinnitus, and cognitive abnormalities. Other, less common, manifestations include coma, Parkinsonism, dementia, and ataxia. Cerebellar hemosiderin deposits [7] and even superficial siderosis [8••] have been noted in some patients with spontaneous intracranial hypotension, particularly in patients with large intraspinal ventral CSF collections of long duration. Spinal manifestations, including myelopathy and radiculopathy, also have been reported and are generally due to compression by large intraspinal ventral CSF collections [9].

The spectrum of clinical and radiographic manifestations of spontaneous intracranial hypotension is unusually broad and many patients with documented spontaneous intracranial hypotension do not meet the diagnostic criteria according to the current International Classification of Headache Disorders (ICHD-2) [10]. Therefore, other sets of criteria have been developed to aid physicians with establishing the diagnosis of spontaneous intracranial hypotension [11, 12]. One area of controversy is the presence of spinal meningeal diverticula without associated CSF leak in patients with spontaneous intracranial hypotension. This was originally included as a criterion in a study published in 2008 [11] but this criterion has since been removed [12], because this finding is not specific to patients with spontaneous intracranial hypotension [13]. The presence of spinal meningeal diverticula also is not included in the new soon to be published criteria of the ICHD (ICHD-3). Kranz and colleagues [14] confirmed that meningeal diverticula are not rare in patients without spontaneous intracranial hypotension and reported diverticula in 44 % of controls vs 68 % in patients with spontaneous intracranial hypotension, while the mean number of diverticula was 2.2 in controls and 6.3 in patients with spontaneous intracranial hypotension. These differences were not statistically significant, but the small number of patients in this study (n = 19) makes a type II error likely.

Although spontaneous resolution of spontaneous intracranial hypotension probably is frequent and placement of a lumbar epidural blood patch is a well-known and reasonably effective treatment for spontaneous intracranial hypotension [15, 16], the management of patients with spontaneous intracranial hypotension is not always that straightforward. A variety of treatments may be necessary to adequately control symptoms, such as multiple or directed epidural blood patching, intrathecal saline infusion, percutaneous placement of fibrin sealant, and surgical repair of the underlying CSF leak. Imaging of the spine is not necessary for most patients with spontaneous intracranial hypotension who are cured with routine lumbar epidural blood patching. However, for directed epidural blood patching, percutaneous fibrin glue placement, and surgical CSF leak repair it becomes imperative to image the spine and detect the exact site of the CSF leak.

CT-myelography has long been considered to be the gold standard of CSF leak detection, especially compared with radionuclide cisternography and routine spine MRI. In 1 study, for example, only two-thirds of leaks detected on CT-myelography could be visualized on radionuclide cisternography or routine spine MRI [17]. However, heavily T2-weighted MRI (or, MR-myelography) has been shown to be an excellent noninvasive alternative to CT-myelography. Wang and colleagues [18] studied 19 patients with spontaneous intracranial hypotension using both modalities and CSF leaks were detected using CT-myelography in 13 patients. MR-myelography showed identical or very similar CSF leaks in these 13 patients but also identified a CSF leak in 2 (33 %) of 6 patients with normal CT-myelogram results. Intrathecal gadolinium-enhanced MRI (or, MR-myelography with intrathecal gadolinium) may also be superior to CT-myelography. Akbar and colleagues [19••] studied 24 patients with suspected spontaneous intracranial hypotension who had normal CT-myelogram results and detected a CSF leak in 5 patients (21 %) who underwent intrathecal gadolinium-enhanced MRI. Spinal subtraction MRI is a novel technique that shows promise in the evaluation of patients with spontaneous intracranial hypotension [20].

Rapid CSF leaks require specialized imaging because by the time CT or MRI is performed following intrathecal contrast administration, the contrast has already spread over many levels and the exact site of the dural tear remains unknown. On CT and MRI, these rapid CSF leaks usually manifests themselves as extensive longitudinal intraspinal extradural fluid collections, almost always ventral to the spinal cord in location. Digital subtraction myelography is the method of choice in detecting the exact level of the dural defect [21, 22, 23••] (Fig. 1). Hoxworth and colleagues have pioneered this technique for ventral CSF leaks [22, 23••]. Contrast is injected intrathecally and digital subtraction acquisition is performed at a rate of 3–6 frames per second. Resolution of these images is high, but susceptible to motion degradation. Dynamic CT-myelography [24], whereby the contrast is injected intrathecally in the CT scanner, is another technique to localize rapid CSF leaks, but I have found digital subtraction myelography to be superior in detecting the actual site of the dural defect.

Fig. 1
figure 1

Left upper: Sagittal T2-weighted MRI shows ventral CSf collections. Right upper: Digital subtraction myelogram reveals exact site of dural defect. Left lower: Intraoperative photograph of dural defect. Right lower: Intraoperative photograph of surgical repair with 7-0 prolene sutures

Full size image

Conclusions

Most patients with spontaneous intracranial hypotension do not need spinal imaging prior to treatment and for those who do, a variety of spinal MRI techniques are available. Rapid spinal CSF leaks, usually treated surgically, are best detected with digital subtraction myelography or dynamic CT-myelography.