Authors: Ajay A. Madhavan (USA), Edward S. Yoon (USA), J. Levi Chazen (USA)
Categories: Neuroimaging and Head & Neck, Spontaneous intracranial hypotension, CSF leak, CSF-venous fistula, Photon counting CT myelogram
Source: Korean Journal of Radiology
Authors: Ajay A. Madhavan, Edward S. Yoon, J. Levi Chazen
Spontaneous intracranial hypotension is a neurologic condition that is caused by a spinal cerebrospinal fluid (CSF) leak. The resulting CSF hypovolemia can manifest as a variety of clinical symptoms, with orthostatic headache being the most common. Although this disease has been recognized for decades, modern understanding of the types of causative spinal CSF leaks, diagnostic imaging tests to localize these leaks, and treatment options has evolved substantially in recent years. In this focused review article, we will provide an overview of the current diagnosis and treatment of spontaneous intracranial hypotension. We will emphasize recent improvements in understanding the pathophysiology of spinal leaks, developments in myelographic techniques to localize CSF leaks, and new treatment options for each type of leak.
Spontaneous intracranial hypotension (SIH) is a neurologic disease that is almost always caused by a cerebrospinal fluid (CSF) leak at the level of the spine. The most common clinical manifestation of SIH is orthostatic headache [1]. However, a wide variety of additional symptoms can occur, including vertigo, tinnitus, cranial neuropathy, and findings mimicking frontotemporal dementia [2].
Most patients with SIH present with abnormalities on contrast-enhanced brain MRI, which can include any combination of dural enhancement/thickening, venous sinus distention, spontaneous subdural hematomas/hygromas, and brain sagging. Quantitative scoring systems can be used to predict the probability of finding a spinal CSF leak, and these are particularly helpful to mitigate subjectivity in assessing for brain sagging [3]. The most common scoring system used for this purpose, frequently referred to as the “Bern score,” assigns points based on the presence of pachymeningeal enhancement, venous sinus engorgement, subdural fluid collections, reduced mammilopontine distance, reduced prepontine cistern distance, and reduced suprasellar cistern size. It can be used to estimate the likelihood of finding a spinal CSF leak on myelography. Importantly, up to 20% of patients with SIH can present with a normal brain MRI findings [4].
An MRI of the spine is also important in patients with suspected SIH, for two main reasons. First, it can detect or exclude the presence of extradural CSF. This includes both well-defined extradural CSF collections, as well as ill-defined CSF signal intermixed with epidural fat [5]. Well-defined extradural CSF collections, sometimes referred to as spinal longitudinal extradural collections, are common in the setting of chronic dural tears. Conversely, dural tears in the acute setting may only present with ill-defined CSF mixed with epidural fat [6]. For the purpose of detecting either, axial and sagittal fast spin echo T2-weighted (T2W) imaging with fat suppression is critical. Furthermore, high-resolution 3D T2W fat suppressed imaging can be added to improve detection of small epidural CSF collections. Second, spine MRI can detect spinal meningeal diverticula (either nerve root sleeve diverticula or arachnoid diverticula), which are associated with some types of spinal CSF leaks [78].
The purpose of this focused review is to describe the current understanding of spontaneous spinal CSF leaks, outline myelographic techniques used to localize these leaks, provide clinical insights for radiologists performing myelography in these patients, and discuss treatment strategies for patients with spinal CSF leaks.
In recent years, the understanding of different spinal CSF leak types causing SIH has improved substantially. Currently, three primary leak types are recognized, including ventral dural tears, lateral dural tears, and CSF-venous fistulas (CVFs) [2].
Ventral dural tears represent physical defects along the ventral surface of the dura, usually caused by sharp osseous spicules or disc-osteophyte complexes (Fig. 1). Importantly, by the time of imaging, the causative osseous spicule may have been resorbed [9].
Lateral dural tears occur most commonly near the axilla of the nerve root sleeves (Fig. 2). They can also occur at the shoulder of the nerve root sleeve or rarely at the level of the pedicle [10]. Lateral tears are frequently associated with herniations of arachnoid that protrude through the dural defect [11]. These arachnoid herniations can sometimes be seen on spine MRI, helping to predict the site of the leak [12].
Finally, spontaneous leaks can be caused by CVFs, which are direct connections between the spinal subarachnoid space and the paraspinal or epidural venous plexus (Fig. 3) [13]. CVFs usually arise in association with nerve root sleeve diverticula [714].
Aside from these three primary leak types, there are less common pathways for spontaneous spinal CSF leaks. For example, CSF leaks into vascular malformations can cause SIH. This typically happens in patients harboring venous or lymphatic malformations in close proximity to the spinal canal [1516]. Additionally, one recent case report described a direct CSF-lymphatic fistula without an intervening lymphatic malformation [17]. It should also be noted that previous publications have classified CSF leaks in slightly different fashions [111819]. In particular, “type 2” leaks have been variably classified as representing lateral dural tears versus leaking meningeal diverticula/distal nerve root sleeve tears. More recent work suggests that the majority of direct lateral epidural leaks likely represent lateral dural tears near the nerve root sleeve. Leaking meningeal diverticula are most likely rare causes of SIH [11].
Brain and spine MRI are crucial initial steps in patients with suspected SIH. The former can both corroborate the diagnosis and exclude other structural pathologies. The latter is needed to determine the type of spinal CSF leak that is likely to be patients with epidural CSF typically have dural tears, while those without epidural CSF usually have CVFs. After initial MR imaging, advanced myelography is needed to detect and localize the spinal leak.
The most important determinant of success during myelography is patient positioning. In nearly all cases, Trendelenburg positioning is required to promote caudocranial contrast flow. If a ventral dural tear is suspected, patients must be prone. If a lateral dural tear or a CVF is anticipated, patients should be decubitus, and this sometimes requires two separate exams to study the left and right side [20]. Additionally, imaging must be performed during or quickly after contrast injection. This technique ensures that dense contrast encounters the site of CSF leak prior to dilution with CSF. Typically, a contrast agent containing 300 mg/mL of iodine is preferred. Generally, when attempting to localize dural tears, relatively small volumes of contrast are used (3–8 mL). When searching for CVFs, larger contrast volumes may be needed (10–20 mL), particularly when performing bilateral decubitus exams and trying to fill meningeal diverticula with dense contrast. Ventral dural tears and CVFs are best seen on early imaging, typically seconds after contrast encounters the site of leak. A small subset of CVFs and ventral tears are more conspicuous on delayed imaging and can require minutes to appear [21]. Lateral dural tears are more variable, and can be seen either immediately or can require a minute or more after contrast injection to become apparent [22].
Numerous imaging modalities are available for myelography, and each has strengths and weaknesses. The first is CT myelography (CTM) using traditional energy integrating detector (EID) scanners [23]. CTM is widely available and accessible. Furthermore, CTM provides cross-sectional imaging with high spatial resolution, usually allowing a slice thickness as low as 0.6 mm. CTM has limited temporal resolution, but this can be enhanced by obtaining multiple successive scans during active contrast injection (referred to as “dynamic” CTM) [24]. Techniques using anywhere from 2–6 scans of the spine have been described for dynamic CTM. These are usually spaced apart by only seconds when assessing dural tears, but can be spaced apart by several minutes for searching for CVFs. Both the spatial and temporal resolution of CTM can be enhanced by using photon-counting detector (PCD) scanners. PCD CTM generates slice thicknesses as low as 0.2 mm and scan speeds are faster compared to EID CTM. These improvements have been shown useful to detect CVFs and dural tears [2526]. PCD CTM also affords the ability to use sharper imaging kernels to maximize spatial resolution, while retaining an acceptable noise level [2728].
Another common modality used for CSF leak detection is digital subtraction myelography (DSM), which is a fluoroscopic technique that employs a subtraction mask immediately before contrast injection. Imaging is then obtained at 1–2 frames per second as contrast ascends in the subarachnoid space [29]. While DSM has excellent spatial and temporal resolution, it does not provide cross-sectional detail. This limitation can be mitigated by performing cone beam CTM immediately after the DSM, which has been shown to improve the yield of this technique for leak detection [30].
Many centers also employ a technique referred to as MR myelography in the work-up of spinal CSF leaks. This terminology has variable definitions in the literature. Some use it to refer to high-resolution or heavily T2W MRI sequences that can be helpful to identify extradural CSF or meningeal diverticula, as described in the previous section. In our experience, these sequences are typically unable to localize CSF leaks. Nonetheless, these sequences are included in our standard MRI protocol and are helpful for identifying subtle extradural fluid collections. By contrast, intrathecal gadolinium MR myelography (GdM) is another imaging modality that has been described for CSF leak detection, which can occasionally localize leaks [3132]. GdM refers to T1-weighted imaging of the spine after instillation of gadolinium contrast into the subarachnoid space. This procedure should be performed with caution, because intrathecal use of gadolinium is off label and exceeding recommended doses can result in neurotoxicity. Typically, only a very small volume (0.1–0.5 mL) of agents such as gadobutrol 1 mmol/mL should be used. Imaging is usually performed within an hour after contrast injection, although delayed imaging can be done to assess for slow leaks [33]. In general, GdM has a lower yield for CSF leak detection compared to CTM and DSM, and therefore we typically use it only in rare cases.
Specific protocols for evaluating patients with SIH vary widely by institution, and further refinements occur continuously. Despite many improvements, spinal CSF leaks are not always localized on a patient’s initial myelogram, even if a good quality exam is done. There are many potential reasons for this.
First, it is possible that the portion of the spine harboring the leak was not initially studied. Most centers first focus their evaluation on the thoracic spine, since this is where the majority of dural tears and CVFs occur [2]. However, both CVFs and dural tears can be seen in other spinal locations [3435]. Therefore, the aforementioned techniques must sometimes be modified to include imaging of the sacrum, cervical spine, or other locations.
Second, even when the entire spine has been evaluated, some leaks can be occult. It may be warranted to repeat myelography in such patients, particular those that have a high pretest probability of SIH. On repeat attempts, radiologists may opt to employ provocative maneuvers such as saline pressurization or resisted inspiration, which have been shown to improve conspicuity of CVFs [3637]. Using a different modality on subsequent studies can help, too. For example, repeat myelograms on PCD scanners can detect subtle fistulas that are missed on lower resolution or non-cross-sectional modalities [38].
Finally, when an initial myelogram is unrevealing, it can be helpful to ensure that the correct type of leak is being sought. For example, dural tears can occasionally present with essentially no epidural CSF on spine imaging, which can confound the picture and lead to missed diagnoses on myelography [39].
Treatment options for spinal CSF leaks depend heavily on the leak type (Fig. 4). For ventral dural tears, a variety of surgical approaches have been employed. Many centers use an open, intradural approach, which permits excellent visualization and direct repair of the dural defect [9]. Recently, endoscopic techniques for ventral dural tear repair have been described, representing a less invasive option [40]. Epidural blood patching can be done as well. Targeted blood patching is preferred to nontargeted approaches. In patients with chronic ventral dural tears causing longitudinally extensive epidural collection, direct fibrin injection into the collection has recently been described [6]. For both applications, CT-guidance is helpful to deliver injectate as close as possible to the site of CSF leak. Contrast injection prior to administration of fibrin or blood is also important to exclude inadvertent vascular or intrathecal injections.
For lateral dural tears, both surgery and epidural blood patching are reasonable treatment options. Surgical closure generally has a higher success rate. Certain types of lateral tears, such as those that are chronic or associated with large arachnoid herniations, may be less amenable to targeted blood and fibrin injection [41].
Finally, for CVFs, available treatment options include surgical ligation, percutaneous fibrin glue injection, and transvenous Onyx embolization. Surgical ligation has a high success rate, and minimally invasive options are available [42]. Fibrin injection and Onyx embolization are also effective strategies with comparable success rates [43]. Both of these techniques require exquisite attention to detail, ensuring that fibrin or Onyx is delivered as close as possible to the nidus of the fistula (Fig. 4). When fibrin injection is performed to treat CVFs, needle placement close to the junction of the diverticulum and draining vein is preferred. However, injection directly into the diverticulum or vein can also be effective. In such cases, careful assessment of the contrast flow pattern is helpful to ensure that intrathecal flow of injectate is minimized.
The diagnosis and treatment of SIH have improved in recent years. In this focused review, we have discussed the current understanding of spinal CSF leak types, advanced myelographic techniques used to detect and localize these leaks, and current treatment options available for each major leak type. Furthermore, we have outlined specific clinical pearls that may be useful to radiologists who may increasingly encounter patients with SIH.