Neuroimaging Findings in Retinal Vasculopathy with Cerebral Leukoencephalopathy and Systemic Manifestations

SUMMARY: Retinal vasculopathy with cerebral leukoencephalopathy and systemic manifestations is caused by TREX1 mutations. High-quality systematic follow-up neuroimaging ﬁ ndings have not been described in presymptomatic and symptomatic mutation carriers. We present MR imaging ﬁ ndings of 29 TREX1 mutation carriers (20 – 65years of age) and follow-up of 17 mutation carriers (30 – 65years of age). Mutation carriers younger than 40years of age showed a notable number of punctate white matter lesions, but scan ﬁ ndings were generally unremarkable. From 40years of age onward, supratentorial lesions developed with long-term contrast enhancement (median, 24months) and diffusion restriction (median, 8months). In these lesions, central susceptibility artifacts developed, at least partly corre-sponding to calci ﬁ cations on available CT scans. Some lesions ( n ¼ 2) additionally showed surrounding edema and mass effect (pseudotu-mors). Cerebellar punctate enhancing lesions developed mainly in individuals older than 50years of age. These typical neuroimaging ﬁ ndings should aid neuroradiologic recognition of retinal vasculopathy with cerebral leukoencephalopathy and systemic manifestations, which may enable early treatment of manifestations of the disease.

nonenhancing white matter lesions (WMLs), 2) WMLs with punctate enhancement, and 3) rim-enhancing lesions with surrounding T2-hyperintensity (edema/gliosis) and/or diffusion restriction. 9 These lesions with surrounding edema may become large with mass effect and are referred to as pseudotumors.
RVCL-S remains a frequently missed diagnosis, partly due to lack of recognition by radiologists. We systematically studied the neuroimaging characteristics of RVCL-S and describe the frequency and evolution of these findings in a large set of presymptomatic and symptomatic carriers.

Participants
All family members 18 years of age and older from 3 unrelated Dutch families with RVCL-S were invited to participate in this study. Family members with unknown TREX1 status were genetically tested. The genetic status of individuals was not disclosed to them, unless specifically requested. The study was approved by the local Medical Ethics Committee. All participants provided written informed consent before inclusion. The study was performed according to guidelines of the Declaration of Helsinki.

Study Design
The study design has been described in a previous publication in which baseline clinical data of the mutation carriers (MCs) was presented. 4 In short, all MCs underwent a baseline MR imaging for research purposes only. Follow-up imaging was acquired only in TREX1 MCs who were aware of their genetic status for combined clinical and research purposes, using the same MR imaging scanner and scan protocol. In addition, CT scans acquired for clinical purposes during the follow-up period were included in the evaluation. The 3T MR imaging scanning protocol consisted of pre-and postgadolinium 3D-T1-weighted, 3D-FLAIR, DWI, SWI, and T2-weighted scans (Online Supplemental Data).

Measurements
The age of MCs was rounded to the nearest half-decade to protect anonymity.
General features of small-vessel disease were evaluated on baseline scans using predefined criteria (Online Supplemental Data).
As earlier described, active RVCL-S lesions were those with punctiform, linear, or rim gadolinium (Gd) enhancement and/or diffusion restriction. 9 Number, location, and, if available, behavior across time were recorded. Colocalization of susceptibility artifacts on SWI was recorded. At follow-up, the duration of enhancement and diffusion restriction of lesions was assessed, defined as visible on at least 2 consecutive time points.

RESULTS
Baseline MR images were available in 29 MCs (17 women) with a median age of 51 years (range, 20-65 years). Follow-up MR images were available in 17 MCs (12 women) with a median age of 53 years (range, 30-65 years). The median follow-up time between baseline and the last MR imaging was 29 months (range, 20-36 months) with a median of 2 follow-up scans per MC (range, 1-7 MR images). For 2 MCs, additional CT scans were available.

Cross-Sectional Neuroimaging Characteristics
Baseline characteristics of small-vessel disease are presented in Table 1. Already, 45% of the MCs younger than  40 years of age showed punctate WMLs, more than expected for their age. In MCs 40 years of age or older, various degrees of white matter involvement were documented and most WMLs were located in the supratentorial periventricular and deep white matter. In contrast to earlier reports, 3,9 the corpus callosum was also affected in a few cases with extensive WMLs. In 1 case, WMLs even seemed to have a characteristic Dawson fingers pattern (Fig 1). This MC did not have a clinical diagnosis of multiple sclerosis; however, a radiologically isolated syndrome of multiple sclerosis was not fully excluded because no CSF analyses or spinal MR images were obtained. In total, 10 MCs showed $1 supratentorial active lesion at baseline; the youngest was 40 years of age. In these 10 MCs, a total of 44 active supratentorial RVCL-S lesions were identified. Their neuroimaging characteristics are presented in Table 2. All basal ganglia lesions showed punctiform enhancement (Fig 2). Linear or rimenhancing lesions were mostly next to the frontal (n ¼ 6) or dorsal horns (n ¼ 5) of the lateral ventricles. Diffusion restriction was observed mainly in the center of lesions, except in a few larger lesions (n ¼ 2) in which restriction was at the periphery. Central punctiform or linear susceptibility artifacts were noted in 68% of lesions (Fig 1). All rim-enhancing lesions showed diffusion restriction and susceptibility artifacts. Although 1 rim-enhancing lesion had some associated edema, no pseudotumors were present at baseline. At baseline, 20 MCs were aware of their mutation status and were tested for vascular retinopathy. Nineteen MCs had signs of retinopathy (mean age, 53 [SD, 8] years). Of these, 10 (53%) had $1 supratentorial active lesion. The MC without retinopathy (50 years of age) did not have active lesions. Due to ethical concerns, MCs who did not wish to know their mutation status could not be tested for retinopathy (n ¼ 9; mean age, 26 [SD, 5] years). None of them had active lesions. At baseline, 11/29 MCs (38%) had features of focal or global brain dysfunction, and 6/29 MCs (21%) had internal organ dysfunction (Online Supplemental Data).

Follow-up Neuroimaging Characteristics
In 5 MCs (4 women; median age, 52 years; range, 30-60 years), no active supratentorial lesions were observed at all, and 1 MC did not develop new lesions during follow-up. Of these, 4/6 MCs had signs of vascular retinopathy. In 11/17 (65%) MCs, a total of 28 new supratentorial lesions developed (Online Supplemental Data). All 11 MCs had signs of vascular retinopathy. During follow-up, 11/17 MCs (65%) had features of focal or global brain dysfunction and 8/17 MCs (47%) had internal organ dysfunction (Online Supplemental Data). We observed long-term enhancement in 46/72 (64%) lesions and long-term diffusion restriction in 26/72 (36%) lesions. In these lesions, contrast enhancement persisted for a median of 24 months (range, 2-37 months), and diffusion restriction, for 8 months (range, 3-32 months). Figure 3 illustrates long-term enhancement and diffusion restriction during 31 months of followup. Most lesions appeared at a certain time point, remained unchanged for a period of time, and eventually became smaller and disappeared with or without a residual lesion of gliosis and/or SWI artifacts. Six lesions were visually documented to increase in size.
Almost exclusively, lesions located in the deep and periventricular white matter became linear or rim-enhancing lesions, while the enhancement of lesions in the basal ganglia remained punctiform. Two MCs (12%) developed a rim-enhancing mass lesion with surrounding edema (pseudotumor) during follow-up (Figs 4 and 5). Both MCs received corticosteroid treatment after the first evidence of the pseudotumor, after which the pseudotumor gradually diminished in size (Fig 5). In 1 of these MCs, we noted, on consecutively acquired CT scans, the development of a punctiform calcification corresponding to SWI artifacts as long as 31 months after the first evidence of the

DISCUSSION
This study systematically reports neuroimaging findings in a group of MCs with RVCL-S in a broad age range, enabling us to describe the presentation and natural history of both subtle and more obvious MR imaging features of the disease. Recognition of these imaging features and the evolution of lesions with time may help to diagnose RVCL-S earlier.
In MCs younger than 40 years, who were mostly clinically asymptomatic, 45% had nonspecific supratentorial punctiform WMLs, remarkable for this age. One recent study also reported WMLs in 5 patients with RVCL-S younger than 40 years of age with comparable WML volumes. 17 Nonspecific WMLs become more apparent in individuals older than 40 years of age. While some MCs still have only discrete punctate WMLs, some develop more extensive confluent WMLs.
Starting from 40 years of age, supratentorial "active" lesions (with contrast enhancement and/or diffusion restriction) may develop and become more prevalent with increasing age. These active lesions were present in 50% of MCs in the 40 years and older age group. Most lesions were relatively small (6 4 mm) and were equally distributed in the periventricular white matter, deep white matter, and basal ganglia, with rim-enhancing lesions mainly found in the deep and periventricular white matter. Before, only punctiform and rim-enhancing lesions were described as characteristic of RVCL-S; 9 however linear enhancing lesions can also be observed.
Follow-up MR imaging demonstrated that many active lesions remain stable or decrease in size and may show long-term persistence of enhancement and diffusion restriction. A general pattern is that RVCL-S lesions first show contrast enhancement as well as diffusion restriction, and while the diffusion restriction fades away after a few months, the contrast enhancement can persist for up to 2 years. This is in accordance with a recent report in 6 symptomatic patients with RVCL-S, in which lesions showed a mean duration of diffusion restriction of 5 months and contrast enhancement of 20 months. 16 Why RVCL-S lesions show such long-term enhancement and diffusion restriction is not fully understood. Histopathologic findings in RVCL-S lesions show focal areas of tissue inflammation and necrosis, which resemble findings in delayed radiation necrosis. In RVCL-S, this is assumed to result from endothelial dysfunction with bloodbrain barrier dysfunction and chronic ischemia. 9 The chronic nature of these focal areas of tissue inflammation and necrosis in RVCL-S may explain the long-term enhancement and diffusion restriction. Few lesions grew during follow-up, and in some lesions, a slow "migration" was observed, leaving earlier affected tissue behind as parenchymal loss/gliosis (Fig 3). In 2 cases, a pseudotumor developed, which is much less frequent than previously reported. 9 This difference might be related to the younger age and fewer symptomatic cases in our study compared with previous literature.
RVCL-S was associated with multiple, bilateral, mostly punctiform enhancing cerebellar lesions, some with central susceptibility artifacts. Infratentorial lesions were also reported in a previous study. 16 A novel finding in this study is that these lesions start to develop with more advanced disease, around 50 years of age. Cerebellar lesions tend to progress in number, not so much in size. Only a few solitary cerebellar lesions grew in size with rim enhancement.
Up to 68% of active RVCL-S lesions were associated with susceptibility artifacts on the SWI scan, which may relate to extravasated blood products (hemosiderin), calcifications, or prominent vascular (venous) structures. Previously, RVCL-S was shown to be associated with parenchymal calcification. 9,15 In Fig 5, we show that focal calcifications may develop some time after the formation of an active RVCL-S lesion, in this case, as long as 31 months after the first evidence of the RVCL-S lesion. Calcifications may occur as dystrophic changes in an area of chronic parenchyma damage or as a result of vessel injury. This possibility was shown in a recent histopathologic study, in which focal calcifications associated with WMLs and granular calcifications in the walls of several vessels were described. 18 In the current study, punctiform SWI artifacts were also seen in the basal ganglia and cerebellum. Proof of calcifications in the basal ganglia and cerebellar hemispheres would be a new finding in RVCL-S. 9 In the past, neuroimaging findings of RVCL-S have been mistaken for multiple sclerosis, vasculitis, or neoplasms, and unnecessary brain biopsies have been performed in some cases. 7,11 Diffusion restriction may be falsely interpreted as a sign of ischemia or infarction, and the features of disrupted blood-brain barrier somewhat resemble late radiotherapeutic effects and necrosis. However, the long-lasting contrast enhancement and diffusion restriction of RVCL-S lesions have not been reported in any other brain disease.

CONCLUSIONS
Although some neuroimaging characteristics in RVCL-S may be observed in other small-vessel diseases as well, the pattern and evolution of neuroimaging findings are typical for RVCL-S. Mainly the long-term contrast enhancement with accompanying long-term diffusion restriction of RVCL-S lesions characterizes the disease. These findings, together with clinical features, should alert radiologists to consider the diagnosis of RVCL-S and enable early treatment of manifestations of the disease. There were no abnormalities on the SWI. After 10 months (B), the lesion grows with new compression due to edema. Diffusion restriction is still noted, but no SWI abnormalities. After corticosteroid treatment, the enhancing lesion slowly diminished in size, and diffusion restriction disappeared at 31 months (C). A new SWI artefact in the center of the lesion is now noted (white arrow), corresponding to a focal calcification on CT acquired at 32 months, while at 15 months, no calcifications were present (Online Supplemental Data).