Subcortical Cystic Lesions within the Anterior Superior Temporal Gyrus: A Newly Recognized Characteristic Location for Dilated Perivascular Spaces

Discussion

Perivascular spaces are leptomeningeal-lined spaces that surround penetrating arteries as they course within the brain parenchyma. The leptomeningeal layers that line these spaces create a small, fluid-filled space around the arterial wall that is distinct from the subpial and SAS, allowing for a slightly different fluid composition than that present in CSF. These structures are usually identified in normal, healthy patients on high-resolution MR imaging and often demonstrate a curvilinear morphologic pattern along the trajectory of penetrating vessels.²

Dilated perivascular spaces are benign expansions of these normal structures, typically located at the inferior basal ganglia along the anterior commissure.³ Giant perivascular spaces, measuring 1.5 cm, most commonly occur within the mesencephalothalamic region, where they may manifest with obstructive hydrocephalus.¹ Although less common, these lesions have been reported to exist focally elsewhere in the brain parenchyma, including the subcortical white matter.¹ Regardless of specific location, enlarged perivascular spaces have been noted to abut either the ventricular margin or the SAS, as seen in the cases of all of our patients.¹ Diffusely enlarged perivascular spaces have also been reported, which have been hypothesized to occur in the context of certain pathologic conditions, including chronic microvascular ischemia, Alzheimer dementia, cerebral amyloid angiopathy, CADASIL (cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy), and mucopolysaccharidoses.^{4⇓⇓⇓–8}

Regarding the function of perivascular spaces, several theories have been proposed. They are thought to play a role in immunologic processes in the brain, to provide a drainage route for interstitial fluid, and to provide a mechanism for equalization of intracranial pressure.⁹ However, the cause of their enlargement, and the reason for regional predilection to dilation, is poorly understood. Proposed mechanisms of enlargement include those related to the associated vessels, namely arterial elongation or increased vascular permeability; those related to the surrounding brain parenchyma, such as volume loss with ex vacuo enlargement; or those related to CSF or interstitial fluid dynamics, including altered or increased CSF pulsations with increased pressure within the perivascular space, or obstruction to flow of interstitial fluid on the basis of amyloid deposition.¹ However, the mechanisms by which these processes may occur focally in some cases, and diffusely in others, remain unexplained.

Previous literature on dilated perivascular spaces has demonstrated the tendency of these lesions to exhibit a regional preference. The largest series of dilated perivascular spaces, reported by Salzman et al,¹ described 37 cases, of which 21 (57%) were located in the mesencephalothalamic region, whereas only 8 (22%) were within the subcortical white matter. It is interesting to note that of all lesions located within the white matter, 50% exhibited surrounding perilesional signal change on FLAIR/T2 sequences, but none of the mesencephalothalamic lesions demonstrated this associated finding. This supports our findings because most of our cases demonstrated perilesional signal alteration. However, Salzman et al¹ did not describe a temporal predominance of the subcortical lesions because only 2 of the white matter lesions were identified within the temporal lobe. A smaller series reported by Cerase et al¹⁰ described 3 cases of dilated perivascular spaces where the lesions regressed with time: All were located in the anterior superior temporal lobe, corresponding exactly to the location we describe in our series, and one of the lesions exhibited perilesional signal change on FLAIR sequence.

The cause of perilesional signal change has also been discussed previously, with a proposed theory being that of accelerated white matter ischemic change resulting from compression of the adjacent parenchymal vessels by the enlarged perivascular space.¹¹ A more interesting theory, however, and one that is reflected in our MR imaging and pathologic findings, is that the high signal may be the result of gliosis in addition to multiple prominent perivascular spaces that are below the resolution of MR imaging.^1,12 Although we were able to prospectively identify prominent perivascular spaces surrounding the dominant lesion in 2 patients only (cases 6 and 11), our pathologic specimen was far more revealing. It is interesting to note that our patient (case 7) underwent surgical excision partially on the basis of MR imaging findings that suggested perilesional vasogenic edema, which was demonstrated on pathologic findings to be nothing more than the presence of multiple prominent perivascular spaces amid diffuse gliosis.

The ultimate objective of studying perivascular spaces is to elucidate imaging features that may help distinguish these benign lesions from more ominous pathologic conditions in a prospective manner. The predilection for certain locations is an important distinguishing feature of dilated perivascular spaces. Although previously only reported to favor the mesencephalothalamic region, our findings, in conjunction with several other reported cases, propose that the anterior superior temporal subcortical white matter is an additional preferential site for dilated perivascular spaces. Other features may also be helpful in identification of these structures, including their orientation and shape, which may often be elongated or fusiform in morphologic appearance, suggesting their association with a penetrating vessel.¹³ In our cases, the proximity to the SAS with the identification of the adjacent middle cerebral artery is an additional attribute that helps to distinguish these lesions, such that they may be associated with the course of a perforating vessel. Finally, the lack of associated enhancement or restricted diffusion, as well as the stability of these lesions with time, also act as key indicators of the underlying pathologic condition. Our cases, as well as those reported in the literature, demonstrate that perilesional FLAIR/T2 signal change may be identified in the context of dilated perivascular spaces and should not, in itself, prompt alternative diagnoses to be entertained.

It is important to note that only 5 of our patients had adequate imaging follow-up (range, 22–112 months) to indicate stability with time. Four patients had follow-up ranging from 6–14 months, and 5 patients did not undergo imaging follow-up at our institution (1 patient [case 7] without follow-up underwent surgery). For the cases that did not undergo imaging follow-up, one cannot rely on imaging stability to aid in diagnosis; however, given that these patients were asymptomatic regarding the lesion, and all other suggestive imaging features were present, including location within the anterior superior temporal lobe, the diagnosis of dilated perivascular spaces can be strongly suspected. In these situations in which the diagnosis of dilated perivascular spaces is highly favored but not proven, surgical management may still be avoided in favor of a more conservative approach.

Other benign lesions in the differential diagnosis that may have a similar imaging appearance include neuroglial or glioependymal cysts, which include the subtype of ependymal cysts. These rare lesions are thought to develop from sequestered embryologic remnants of neuroectoderm or neural tube elements.¹⁴ Although they demonstrate regional preferences—ependymal cysts are commonly identified in a juxtaventricular location, whereas other neuroglial cysts demonstrate frontal lobe predominance—these locations do not coincide with that of dilated perivascular spaces, thereby providing another basis on which to distinguish these pathologic conditions.^14,15 Acquired lesions, on the other hand, including porencephalic cysts, cystic encephalomalacia, or sequelae of lacunar infarctions, would not be expected to demonstrate a regional preference, given that they are the result of prior insults to the brain that would not necessarily occur in a specific location. Finally, infectious cysts and cystic neoplasms would not necessarily follow a CSF signal on all sequences, could demonstrate associated enhancement, and would be expected to enlarge or change with time.