Tumefactive Primary Central Nervous System Vasculitis: Imaging Findings of a Rare and Underrecognized Neuroinflammatory Disease

DISCUSSION

Our study describes and illustrates the spectrum of imaging findings of t-PCNSV, which is a rare disorder manifesting as masslike brain lesions. T-PCNSV seems to preferentially involve small parenchymal and leptomeningeal vessels, usually sparing large to medium-sized intracranial vessels, which may appear unremarkable on commonly performed imaging studies such as CTA or MRA.

Radiologic features of t-PCNSV were not well described or recognized because it is extremely rare entity.⁴ To our knowledge, apart from several case reports^11-22 and review articles,^23⇓-25 only a few small case series have been published in the medical literature.^6⇓-8,10 Molloy et al⁷ reported 8 cases of masslike PCNSV with the proved histopathology including both “Aβ-related” PCNSV and “non-Aβ-related” PCNSV from 2 tertiary institutions. In addition, they included a literature review of 30 cases. Lee et al¹⁰ reported 4 cases of biopsy-proved lymphocytic PCNSV mimicking brain tumor on MR imaging. de Boysson et al⁸ described 10 masslike PCNSV cases in their series. Nine cases had histopathology-confirmed PCNSV; 1 case had histopathology negative for PCNSV but had large and medium vessel abnormalities suggestive of PCNSV on both MRA and DSA. In their series, they did not specify the absence or presence of Aβ on histopathology. Recently, Salvarani et al⁶ have reported a comprehensive analysis of 13 biopsy-proved cases of masslike PCNSV; however, 6 were found to have Aβ-related PCNSV (or ABRA/CAA-RI), leaving only 7 non-Aβ-related PCNSV cases. None of these case series identified definitive radiologic features of t-PCNSV. Furthermore, the inclusion of Aβ-related PCNSV or ABRA/CAA-RI in their sample size may preclude their analysis of the radiologic features of non-Aβ-related PCNSV. In contrast, our study is exclusively focused on neuroimaging features of t-PCNSV with histopathologic correlation and we did not incorporate Aβ-related PCNSV or ABRA/CAA-RI in our analysis because these have recognized characteristic imaging patterns described in prior literature.^27⇓⇓-30

Our t-PCNSV cohort had a male predominance, which was observed in previous studies.^3,6,8,25 In our cohort, the most common presentations of patients with t-PCNSV were headaches, seizures, and focal neurologic deficits, which were similar to prior studies.^6⇓-8,25

In our study, even though-PCNSV had a variety of imaging patterns, we could identify several imaging features shared by many of the patients with t-PCNSV, which may help raise the possible diagnosis of this disease. We found that in t-PCNSV, a single lesion was more common than multiple lesions. The most common location of t-PCNSV was the supratentorial cortex/subcortical areas, as shown in previous studies.²⁵ Deep gray nuclei involvement can be seen occasionally. Only 1 patient had brain stem involvement. All patients with t-PCNSV in our cohort demonstrated enhancement, which had various patterns. Patchy and nodular enhancement patterns were the most common. Less common were ring and linear/perivascular enhancement, respectively. Some patients who demonstrated patchy enhancement had multifocal small hypoenhancing areas within a background of patchy enhancement, resembling a “mottled appearance” (Fig 1A–C). We hypothesized that this unique pattern of enhancement may represent superimposed small infarctions or necrosis within the background of brain parenchymal inflammatory processes. A minority of the patients showed subependymal or leptomeningeal enhancement associated with the main lesions, thought to represent inflammatory processes involving adjacent leptomeningeal vessels and subependymal vessels, respectively.

Interestingly, in our study, we found that none of the patients with t-PCNSV demonstrated diffusion restriction on DWI (Fig 3B), which several previous studies have reported,^6,8,10 despite frequent presence of microinfarctions on histopathology.²⁶ We could hypothesize that this may reflect the involvement of very small parenchymal/leptomeningeal vessels (including arterioles, capillaries, and/or venules) found in combination with extensive brain parenchymal inflammation, in contrast to medium or large vessel occlusions causing “bland” acute infarctions (54%–85%).^4,9

In our study, we found that most patients with t-PCNSV demonstrated microhemorrhages within the lesions, which can be detected on either GRE-T2* or SWI (Figs 1E–H). The microhemorrhages can appear either punctate or linear. The pathophysiology of parenchymal microhemorrhages in t-PCNSV is not well understood; one proposed mechanism is the alteration of vessel wall permeability secondary to the vasculitis process itself.⁹

A minority of our patients with t-PCNSV had available advanced MR imaging. Three patients underwent MRP (Fig 3E). There was a low rCBV value that could be due to a lack of neoangiogenesis unlike malignant brain neoplasms, concurring with previous case reports on MRP of t-PCNSV.^14,17,20 Two patients underwent MR spectroscopy, which showed a nonspecific pattern and did not add specific findings helpful in the exclusion of neoplasm from the diagnosis.^{10,13,14,18,21,25} Three patients had normal vessel wall MR imaging of the large arteries in the circle of Willis.

Our study showed that most t-PCNSV was lymphocytic vasculitis (90%), while only 10% was granulomatous vasculitis. This corresponds to some previous studies.^8,10,25 However, other previous studies^6,7 reported that granulomatous vasculitis was more common in PCNSV with a tumorlike presentation. Those studies^6,7 included patients with ABRA/CAA-RI, which commonly have granulomatous vasculitis and Aβ deposits in the vessel wall.

In our study, we did not include the patient who was diagnosed with ABRA/CAA-RI. Some previous studies^{1,5⇓-7,27,30} have classified ABRA/CAA-RI as a specific subgroup of t-PCNSV. Although ABRA/CAA-RI share similar patterns of angiocentric inflammatory cell infiltrates of vessel walls, the pathophysiologic mechanism is different as it involves Aβ deposition in the walls of small-to-medium vessels mostly arteries in the leptomeninges and cerebral cortex.^5,27 It is not fully understood whether the deposition of Aβ in vessel walls incites the inflammatory response or, conversely, if the inflammation leads to Aβ deposition. These entities have characteristic imaging patterns described as lobar subcortical edema and cortical/subcortical microhemorrhages associated with predominant leptomeningeal enhancement.^27⇓⇓-30 ABRA/CAA-RI rarely show parenchymal enhancement seen in our patients with t-PCNSV. The pattern of microhemorrhage is also usually more diffuse in ABRA/CAA-RI as compared with our patients with t-PCNSV, where it was preferentially confined to the areas of signal abnormality and enhancement. An additional reason to exclude ABRA/CAA-RI cases is that ABRA/CAA-RI were commonly seen in older men and those who have other comorbid illnesses and tend to have poorer outcomes compared with t-PCNSV. Furthermore, previous studies in some patients with Alzheimer disease who received amyloid-modifying therapies were found to have amyloid-related imaging abnormalities, which resemble ABRA/CAA-RI on MR imaging.^31,32 Its potential mechanism is that immunotherapies cause an increase of clearance of parenchymal plaque Aβ with a transient increase in vascular Aβ, leading to temporary vascular leakage. This supports the hypothesis that Aβ plays an important role in the pathomechanism of ABRA/CAA-RI.^31,32 Therefore, we postulated that ABRA/CAA-RI, where Aβ deposits are presumed to be the causative factor, could represent an entity different from t-PCNSV, where these deposits are absent.

Differentiation between t-PCNSV and brain neoplasms on imaging alone can be challenging. Advanced imaging techniques such as MRP and MR spectroscopy may aid in the diagnosis. t-PCNSV does not show increased relative CBV, which is a feature of neoangiogenesis seen in high-grade gliomas. Differential diagnosis based on imaging findings would often include primary CNS lymphoma (PCNSL). Intralesional microhemorrhages are not an uncommon imaging feature of PCNSL.³³ Furthermore, a previous study has shown that systemic lymphoma can occur in about 6% of patients with PCNSV.³⁴ These patients may have a prior history of systemic lymphoma or simultaneous systemic lymphoma detected on PCNSV work-up.³⁴ Radiologically, the lack of diffusion restriction may be helpful in differentiating t-PCNSV from PCNSL. In our cohort, none of our patients had systemic lymphoma, either at the time of the biopsy or on follow-up. The presence of intralesional microhemorrhages and lack of specific open ring enhancement or peripheral diffusion restriction assisted in differentiating t-PCNSV from tumefactive demyelinating lesions.³⁵ Other unusual atypical infectious conditions such as CNS tuberculosis and fungal infections should also be included in the differential diagnosis of t-PCNSV.

Our study has several limitations; first, as a retrospective study the imaging protocols applied were heterogeneous and some patients did not have complete laboratory tests and follow-up. Second, the extreme rarity of t-PCNSV prevented the undertaking of large-scale studies. We collected cases from different institutions, which may have caused a selection bias. Third, PCNSV has a different disease subtype and multifaceted clinical and histopathologic appearances resulting in difficulty in the grouping of imaging appearances for the diagnosis. Furthermore, we did not have available DSA, which is considered the reference standard for the preoperative diagnosis of PCNSV, because those patients were thought to have tumors preoperatively. Once the histopathologic diagnosis of PCNSV has been established, DSA is usually not deemed necessary especially in the era of CTA and MRA. It would have been helpful if there were available DSA, though DSA is often negative if the affected vessel diameter is less than 200 µm in diameter.³⁶ In addition, previous studies have already shown that most patients in this subgroup have normal DSA findings.^4,6⇓-8