Dr. Krishnamoorthy Thamburaj
From the days of limited knowledge through the autopsy series, imaging techniques have vastly expanded our understanding of almost every aspect of cerebral venous thrombosis (CVT), including demography, pathophysiology, clinical manifestations, treatment options, and outcome. CVT is responsible for only 0.5 to 1% of stroke; however, it is well known to pose a considerable challenge to diagnosis on clinical assessment and imaging.1 It is no surprise, then, that results from the largest cohort of CVT cases demonstrated a median delay of 4 days in admission to a hospital and 7 days in diagnosing CVT from the onset of symptoms.2 Fortunately, CVT carries a better prognosis in approximately 85% of cases and deep venous thrombosis is no exception.2,3 CVT carries mortality and significant morbidity in 13–15% of cases, which intensifies the need to recognize the condition at the initial presentation, initiate appropriate treatment measures, and improve the outcome.2
Anticipation of the diagnosis and a standard approach to analyze the cerebral venous structures on routine imaging are essential steps for a successful identification of CVT. Analysis should focus on recognizing as well as searching for connections between the direct signs of venous clot and the indirect signs of parenchymal changes, features of raised intracranial pressure, venous collaterals, dural thickening, etc. In emergent situations, the diagnosis of CVT is often unsuspected clinically and the patients typically undergo a head CT. Depending on the time of clinical presentation and location of the venous clot, noncontrast CT may demonstrate several imaging features, including hyperattenuating venous clots (dense delta sign, hyperattenuating sinus, dense dural sinus, dense clot sign, and cord sign), parenchymal changes of swelling, edema, hemorrhage, brain shifts, and occasionally subarachnoid hemorrhage, mainly in the convexity sulci.4 Advances in multisection CT technology have made it possible to identify the venous clots with noncontrast CT in more than 70% of CVT cases, especially when the diagnosis is anticipated at the time of image analysis.5 The ability to obtain thinner CT sections and multiplanar reconstructions can be expected to improve the diagnosis of CVT in relatively difficult locations such as transverse and sigmoid sinuses. Indeed, transverse and sigmoid sinuses are either the most common or second most common locations for CVT.2 Obtaining contrast-enhanced head CT, though not a common practice in the emergency setting, can further improve the sensitivity of CT and aid in identifying the signs of venous clot, such as empty delta sign.4 A word of caution is important at this stage in analyzing noncontrast CT, because false-positive venous clot may occur from high hematocrit and false negative clot from low hematocrit, as well as delayed clinical presentation after a week or two. Quantitative assessment of Hounsfield units (HU) in the venous sinuses and their relationship to hematocrit can be expected to improve the diagnosis of CVT. Black et al6 reported a positive correlation between hematocrit and Hounsfield values in dural sinuses. Besachio et al7 proposed a cutoff value of >65 HU and a Hounsfield unit to hematocrit ratio of (H:H) >1.7 to identify venous clots. Buyck et al8 reported that an optimal threshold of ≥62 HU has 95% accuracy and an H:H ratio of ≥1.52 has 97.5% accuracy when detecting CVT.
Due to the unique ability to alter the magnetic signal, evolutionary changes of hemoglobin provide superior capability to MRI to not only detect a venous clot but also identify its approximate age at the time of presentation.4,9 Many reviews presume MRI is superior to CT to diagnose CVT.1,10,11 Undoubtedly, subacute venous clots are best identified on MRI, as they lose attenuation on noncontrast CT and appear hyperintense on MRI due to the influence of methemoglobin in the clot. With significant progress made in the field of imaging technologies, a direct comparison between CT and MRI under identical clinical background and time of imaging is necessary in order to learn the sensitivity, specificity, accuracy, and benefits of the diagnosis of CVT, particularly in the acute stage. In a recent study, Patel et al12 reported that routine brain MRI sequences identified dural sinus clots in 80% of cases of CVT. The authors correctly stressed the importance of analyzing a combination of MR sequences to identify venous clots, because false-negative and false-positive CVT are known to occur with MRI as in CT. False-positive clot may occur from flow-related phenomena in sinuses, resulting in bright intraluminal T1 signals, and false-negative clot on T2 sequences from signal loss mimicking patent flow void due to deoxyhemoglobin or intracellular methemoglobin. In their study, moderate interobserver agreement was noted in identifying venous clots, which highlights the continuing diagnostic challenge presented by CVT.12 T2* gradient-echo sequence (GRE) is considered the best MR sequence to identify a venous clot in the acute stage, which usually lasts from 3 to 5 days after the onset of CVT.13 Isolated cortical vein thrombosis is responsible for approximately 6% of CVT, and it is also best identified on a T2* GRE sequence.14,15 Depending on the area of drainage, it can result in a large space occupying venous infarct. With the introduction of high-resolution susceptibility-weighted imaging and its variants in clinical service, we should expect to see further progress in the diagnosis and management of CVT.16 Venous clot will appear as a filling defect in a contrast-enhanced T1 sequence. An awareness of the anatomic variants such as early bifurcation of superior sagittal sinus, intrasinus septations, and arachnoid granulations is important to avoid a false-positive diagnosis of CVT on contrast-enhanced MRI.4 Occurrence in well-known locations, shape, relationship to a draining vein, and signals almost similar to the CSF in all MR sequences should help identify arachnoid granulations correctly. Extracellular methemoglobin, which appears approximately a week after the onset of CVT, leads to bright signals in a venous clot on all MR sequences. Bright signals against all other structures in the brain make it easier to identify CVT on a diffusion-weighted sequence at this stage.17 Because extracellular methemoglobin loses the susceptibility effects, it becomes unsuitable for detection by susceptibility-weighted sequences. High-resolution postgadolinium 3D T1 volumetric sequences have been reported to be 83% sensitive and almost 100% specific to identify CVT.18 They are also good at identifying the normal venous structures, such as septations and arachnoid granulations. High spatial resolution data from 3D T1 postcontrast images remain to be explored in the identification of cortical venous thrombosis. On 3D T1 contrast-enhanced MRI, hyperintense T1 signals from both forms of methemoglobin may mimic flow and patency, which further highlights the need to pay attention to all MR sequences to identify CVT. Furthermore, chronic thrombosis continues to pose a challenge to the diagnosis of CVT on contrast-enhanced T1 sequences, including the volumetric 3D T1, because the enhancing clot may be mistaken for a patent sinus.18 More recently, Yang et al19 reported the utility of an optimized 3D turbo spin echo sequence (3D SPACE T1 and its variants) to obtain MR black-blood thrombus imaging and volume quantification of venous clots. Future studies may show whether this information can help guide the treatment of CVT.
More than 50% of CVT cases may demonstrate parenchymal changes consisting of edema and hemorrhage.4,20 Involvement of the cortical veins and the status of venous collaterals play an important role in the development of parenchymal changes compared with the location of venous clots in a dural sinus alone.21 Parenchymal edema may take the form of vasogenic edema, cytotoxic edema, or a combination of both. Vasogenic edema is much more commonly observed and it is often reversible with treatment.22 In CVT, rupture of venules from increased venous pressure is thought to be responsible for hemorrhage, which is typically observed in the subcortical white matter and is often multiple. Tsai et al22 used catheters to measure intrasinus pressure ranging from 20 mm Hg to 50 mm Hg in 11 cases of CVT. Patients with severe edema with or without hemorrhage demonstrated pressure measurements of more than 42 mm Hg. Contrast-enhanced MRI may demonstrate gyriform enhancement of the affected parenchyma in 1–29% of cases of CVT, in addition to venous collaterals and dural thickening from venous congestion.4 Knowledge of cerebral venous drainage is essential to identify parenchymal changes and distinguish them from tumor or encephalitis.4 Close scrutiny of the transverse sinus and vein of Labbé is warranted in temporal lobe lesions, the deep venous system in thalamic and periventricular white matter lesions, and the superior sagittal sinus in parasagittal cerebral lesions. Alteration in cerebral hemodynamics in CVT can be assessed with perfusion imaging, but its true potential in the diagnosis and management of CVT remains to be explored.4 MRI is much more sensitive when identifying the parenchymal changes, differentiating vasogenic edema from cytotoxic edema, and guiding treatment decisions.1,4,11
Venography is an essential step in the management of CVT. 2D time of flight (2D TOF) MR venography (MRV) of the head is considered to be the most commonly used technique to diagnose and monitor the treatment response in CVT, although in many institutes, 3D phase-contrast (3D PC) MRV is also frequently used.1,4,10,11 3D PC MRV is more time-consuming, susceptible to turbulence-induced loss of flow signals, and less commonly used to assess cerebral veins.4 Recently, dynamic first-pass bolus contrast-enhanced MR venography and 4D MR venography have been introduced to obtain rapid information on cerebral venous anatomy and avoid the pitfalls of noncontrast MRV techniques, including 2D TOF and 3D PC MRV, in the diagnosis of CVT.23 2D TOF MRV may result in false-positive sinus occlusion due to flow saturation or a developmentally hypoplastic or atretic sinus and false-negative venous clot due to hyperintense signals from extracellular methemoglobin mimicking flow. Leach et al24 reported contrast-enhanced MRV as the best technique to identify chronic venous thrombosis, a condition notoriously known to be missed on contrast-enhanced T1 sequences. Noncontrast MR venography with 2D TOF or 3D PC should be preferred in pregnant women, patients with severe contrast allergy, and those at risk of developing nephrogenic systemic fibrosis as in chronic renal failure. Advances in multisection CT technology have vastly improved CT venography of the head, which provides more spatial resolution than MRV and is considered at least as good as MRV in demonstrating CVT.25,26 Ionizing radiation is one of the main drawbacks of CT venography, especially when bone subtraction technique is used, because it necessitates double scanning with and without intravenous contrast for subtraction.