key: cord-1031896-bxd8etma
authors: Chougar, Lydia; Shor, Natalia; Weiss, Nicolas; Galanaud, Damien; Leclercq, Delphine; Mathon, Bertrand; Belkacem, Samia; Stroër, Sebastian; Burrel, Sonia; Boutolleau, David; Demoule, Alexandre; Rosso, Charlotte; Delorme, Cécile; Seilhean, Danielle; Dormont, Didier; Morawiec, Elise; Raux, Mathieu; Demeret, Sophie; Gerber, Sophie; Trunet, Stéphanie; Similowski, Thomas; Degos, Vincent; Rufat, Pierre; Corvol, Jean-Christophe; Lehéricy, Stéphane; Pyatigorskaya, Nadya
title: Retrospective Observational Study of Brain Magnetic Resonance Imaging Findings in Patients with Acute SARS-CoV-2 Infection and Neurological Manifestations
date: 2020-07-17
journal: Radiology
DOI: 10.1148/radiol.2020202422
sha: 70269c710d990fffcc0852cfad8a5c82b5f04ca2
doc_id: 1031896
cord_uid: bxd8etma

BACKGROUND: This study provides a detailed imaging assessment in a large series of COVID-19 patients with neurological manifestations. PURPOSE: To review the MRI findings associated with acute neurological manifestations in COVID-19 patients. METHODS: This was a cross-sectional study conducted between March 23 and May 7, 2020 at the Pitié-Salpêtrière University Hospital, a reference center for COVID-19 in the Paris area. Inclusion criteria were: adult patients diagnosed with SARS-CoV-2 infection, presenting with acute neurological manifestations and referred for a brain MRI examination. Patients were excluded if they had a previous history of neurological disease. The characteristics and the frequency of different MRI features were investigated. The findings were analyzed separately in patients in intensive care units (ICU) and other departments (non-ICU). RESULTS: During the inclusion period, 1176 consecutive patients were hospitalized for suspected COVID-19. Out of 308 patients with acute neurological symptoms, 73 patients met the inclusion criteria (23.7%) and were included: 35 ICU patients (47.9%) and 38 non-ICU patients (52.1%). The mean age was 58.5 ± 15.6 years, with a male predominance (65.8% vs. 34.2%). Forty-three patients presented pathological MRI findings 2-4 weeks after symptom onset (58.9%), including 17 with acute ischemic infarct (23.3%), 1 with a deep venous thrombosis (1.4%), 8 with multiple microhemorrhages (11.3%), 22 with perfusion abnormalities (47.7%), 3 with restricted diffusion foci within the corpus callosum consistent with cytotoxic lesions of the corpus callosum (CLOCC, 4.1%). Multifocal white matter enhancing lesions were seen in 4 ICU patients (5%). Basal ganglia abnormalities were seen in 4 other patients (5%). The cerebrospinal fluid (CSF) analysis was negative for SARS-CoV-2 in all tested patients (n=39). CONCLUSION: In addition to cerebrovascular lesions, perfusion abnormalities, CLOCC and ICU-related complications, we identified two patterns including white matter enhancing lesions and basal ganglia abnormalities that could be related to SARS-CoV-2 infection.

Since the coronavirus disease 2019 (COVID-19) outbreak in December 2019, there is growing evidence that the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has central nervous system (CNS) involvement in addition to the primary respiratory target. In a series of 214 patients infected with SARS-CoV-2, more than a third of patients had neurological manifestations, such as stroke, seizures and anosmia (1) . Clinical reports (2) (3) (4) and experiments (5, 6) potential. Coronaviruses enter the CNS using a hematogenous pathway or through the olfactory bulb or peripheral nerves (5, 6) . Using mice models, the infection has been shown to start in the respiratory epithelium, then spread to the brain via the olfactory bulb, and gradually invade the subcortical and cortical regions (5) (6) (7) . This olfactory involvement could explain the anosmia observed in a large number of COVID-19 patients. Proposed pathophysiological mechanisms include a direct viral replication and an immune-mediated reaction (5, 6) . It is not yet known whether this knowledge applies to the new SARS-CoV-2. So I n P r e s s far, several imaging findings have been described in the setting of COVID-19, but the relationship with SARS-CoV-2 remains unclear (8) (9) (10) (11) (12) (13) (14) .

This study reports a series of patients with acute neurological manifestations admitted for COVID-19 and referred for a brain MRI in a Parisian reference center for the disease. We describe the MRI findings and their frequency and investigate whether imaging patterns possibly related to COVID-19 could be identified. These patterns were considered as possibly related to COVID-19 when 1) they were found in ≥3 patients, 2) they were similar across patients, and 3) they were not explained by another pathology or condition.

The study was conducted according to good clinical practice, sponsored by Assistance Publique-Hôpitaux de Paris, and received approval from the local Sorbonne University Hospital Ethics Committee (CER-202028 on 24/04/2020). All patients or their relative signed written informed consent for the use of their medical data in accordance with the French regulation and the European General Data Protection Regulation (GDPR). The study was registered in the clinicaltrial.gov website (NCT04362930). Patients referred for brain MRI in the context of COVID-19 between March 23 and May 7, 2020 were assessed in the Neuroradiology department of the Pitié-Salpêtrière Hospital, a tertiary neurology center and reference center for COVID-19. The inclusion criteria were: i) diagnosis of SARS-CoV-2 infection by detection of RNA in a nasopharyngeal swab, bronchial aspiration or bronchoalveolar lavage using reverse-transcriptase-polymerase-chain-reaction (RT-PCR) or by chest computed tomography (CT) showing results consistent with SARS-CoV-2-associated pneumonia, ii) presence of acute neurological symptoms, iii) age greater than 18 years and I n P r e s s iv) availability of a brain MRI. The exclusion criteria were: i) underlying progressive CNS disease excluding stroke, ii) alternative diagnosis assessed during the MRI examination. A case of deep venous thrombosis from the current series has previously been published (15) .

Clinical data collected by expert neurologists, electroencephalogram (EEG), cerebrospinal fluid (CSF) and chest CT findings were retrospectively extracted from electronic medical records. Difficult cases were discussed in multidisciplinary meetings.

Patients were scanned using a 3.0 Tesla MRI system (PREMIER, General Electric Healthcare) with a 48-channel receive head coil. The protocol included three-dimensional (3D) T1weighted (Magnetization-prepared rapid acquisition with gradient-recalled echo, MPRAGE) and 3D Fluid-attenuated inversion recovery (FLAIR) images, susceptibility-weighted imaging (SWI), 3D pseudo-continuous arterial spin labeling (ASL) perfusion imaging, diffusionweighted imaging (DWI) sequence, 3D post-gadolinium spin echo T1-weighted imaging. In case of suspicion of acute stroke, axial FLAIR sequence, DWI, SWI, ASL and arterial 3D timeof-flight were acquired.

Two neuroradiologists (L.C., N.P.) independently analyzed all MRI images. In case of disagreement, images were reviewed and a consensus was reached. The following imaging characteristics were evaluated: diffusion abnormalities, white and gray matter FLAIR signal abnormalities, ASL perfusion abnormalities, parenchymal and meningeal enhancement, I n P r e s s hemorrhages, vessel permeability, cranial nerve abnormalities including thickening, abnormal T2 hyperintensity and contrast enhancement.

The age was expressed as mean and standard deviation. Categorical variables were expressed as counts and percentages. Participants were divided into an "ICU" group, including patients hospitalized or discharged from an ICU, and a "non-ICU" group, with patients never hospitalized in ICU. The age was compared between groups using Student's ttest. Proportions for categorical variables were compared using Pearson's Chi-squared test or Fisher's Exact Test. Statistical analyses were performed with R software (version 3.6.1).

The significance threshold was set at a p<0.05.

During the inclusion period, 1176 consecutive patients were hospitalized for suspected COVID-19 in the Pitié-Salpêtrière Hospital. Three hundred and eight of these patients Table E1 . (Table 2) . Clinical details for patients with white matter lesions and basal ganglia abnormalities are presented in Table E2 .

Twelve of the 17 patients with acute ischemic infarct (70.6%) had multiple ischemic foci while 8 (47.0%) had a territorial infarction (Supplementary Figure E1) . Three patients had both types of lesions. There was no significant difference between non-ICU and ICU patients (p=0.638).

Extensive deep cerebral venous thrombosis complicated by hemorrhagic venous infarction was diagnosed in a 72-year-old male ICU patient without known risk factor for thrombosis and with normal baseline coagulation tests (Table E3 ).

ICU patients had a greater number of multiple microhemorrhages (≥5) than non-ICU patients (20.6% vs. 2.7%, p=0.017), involving the corpus callosum in 5 patients ( Figure 2 ). Patients with microhemorrhages tended to have increased partial thromboplastin time due to preventive anticoagulation therapy (Table E3 ). Anticoagulation overdose occurred in one ICU patient. Two ICU patients with multiple microhemorrhages (28.6%) had an extracorporeal membrane oxygenation (ECMO).

Twenty-two out of 46 patients (47.8%) presented perfusion abnormalities that were related to seizure in nine patients (19.6%), recent or old vascular lesions in four patients (8.7%) and I n P r e s s that were unrelated to seizures or ischemia in 10 patients (21.7%). In two patients, perfusion abnormalities were attributed to both seizures and ischemia. ICU patients had more perfusion abnormalities than non-ICU ones (62.5% vs. 31.8%, p=0.037), with a higher rate of isolated perfusion abnormalities (33.3% vs. 9.1%, p=0.046). In particular, marked post-ictal changes associating edema and diffusion restriction within the right frontobasal cortex were observed in a 69-year-old ICU patient who presented with status epilepticus (Supplementary 

The pattern of white matter enhancing lesions observed in four patients differed from the one seen in acute disseminated encephalomyelitis (ADEM) where lesions tend to be asymmetrical, with variable involvement of gray matter and punctate or ring enhancement (16) (17) (18) . Several features also argued against the diagnosis of subacute embolic infarctions, including i) the periventricular topography of lesions, ii) their morphology associating an oval shape and ill-defined margins, and iii) the perivascular distribution of enhancement. This latter feature has been described in vasculopathies such as PRES and Susac syndrome, and in disorders with angiocentric infiltrates, especially in neurolupus and neurosarcoidosis (19) and in CD8 encephalitis in HIV patients (20). Similarly, our pattern of white matter enhancing lesions could somehow reflect a vasculitis-like phenomenon and/or an inflammatory disorder with angiocentric involvement (19, 20) . This pattern has also been observed in a recent study (9) , reinforcing the hypothesis of a COVID-19-related damage.

Basal ganglia involvement seen in four patients included signal and diffusion abnormalities, with variable contrast enhancement, affecting the substantia nigra, the globus pallidus and the striatonigral pathway in a variable manner. To our knowledge, this pattern has not been previously reported. Patients with hyperglycemia can present spontaneous T1 hyperintensity within the putamen with variable involvement of the caudate and globus pallidus that is usually unilateral (21) . Although patients with T1 hyperintensities in our series were diabetic, the topography that we observed was different (symmetrical and involving the globus pallidus and the upper part of the striatonigral pathways). This pattern of brain damage has some similarities to that seen in encephalitis lethargica (EL) also known as Von Economo

disease. An epidemic of EL occurred during the 1918 influenza pandemic. In the acute phase, patients with EL had pharyngitis, sleepiness, ocular motility and movement disorders with a fatal presentation in about one third of patients. Delayed parkinsonism and neuropsychiatric I n P r e s s signs may occur (22) . Neuropathological studies have shown encephalitis of the midbrain and basal ganglia with lymphocytes infiltration (22) (23) (24) . MRI examinations revealed bilateral edematous changes within the thalami, basal ganglia and midbrain with variable contrast enhancement (22, 23) . The substantia nigra was involved in 12% of autopsy cases (22) and in one MRI case (25) . Although the etiology of EL is still unknown, infectious and environmental causes have been proposed (24) . More recently, examination of archived brain material failed to demonstrate Influenza RNA, while the presence of oligoclonal bands in the CSF and the efficacy of corticosteroids suggested that EL might be immune-mediated (23) . Follow-up of the patients with basal ganglia abnormalities in our series will help show whether they develop parkinsonism as seen in the chronic phase of EL (24) .

The underlying pathophysiology of these two patterns and their association with COVID-19 remain unclear since SARS-CoV-2 RT-PCR in the CSF was negative in all tested patients. A relationship with COVID-19 appears however possible since these lesions were similarly observed in several patients and were not explained by another pathology or condition. To date, SARS-CoV-2 RNA has been reported positive in the CSF in only three previous patients (9, 26, 27) . Several hypotheses could be formulated to explain PCR negativity in the CSF. First, meningeal contrast enhancement was rare in our series, unlike observations from previous studies (8) . This difference could be due to the fact that we acquired the FLAIR images before the administration of contrast (the meningeal contrast enhancement in other cases being reported mainly on post-contrast FLAIR images), or to the lack of meningeal involvement in our series, considering that SARS-CoV-2 could have an intra-neuronal localization as reported with SARS-CoV (28) . Second, indirect viral pathogenesis through an immune-mediated mechanism and/or a systemic inflammatory response syndrome could be involved. Such an immune process could also explain the occurrence of neuritis or Guillain-I n P r e s s Barré syndrome (29) . Third, the delay between the RT-PCR and the stage of infection could also explain a negative detection. Brain damage may also result from a first step of CNS viral replication, followed by an immune-mediated phenomenon (5, 7) where the virus is no longer detectable, as supported by the relatively long delay between the first respiratory symptoms and the neurological impairment in our study. Post-mortem neuropathological examinations in 18 COVID-19 patients only showed hypoxic changes without sign of encephalitis (30) . Viral RNA was detected at low levels, possibly due to in situ viral RNA from the bloodstream (30) . Another neuropathological report evidenced vascular and demyelinating changes in one patient, without mention of virus screening (31) . The knowledge collected at this stage would favor the hypothesis of a delayed immune-mediated process underlying the physiopathology of CNS damage (14) .

Multiple microhemorrhages have been observed in eight patients, with a specific involvement of the corpus callosum, as recently reported (9, 11, 32) . Hemorrhages are known to be a complication of ECMO (33) . Microhemorrhages might have been explained by the presence of ECMO in 2 patients and anticoagulation overdose in another patient. In the remaining patients, the physiopathology could involve microvascular damage. A possible implication of SARS-CoV-2 remains unclear.

Ischemic infarcts were diagnosed in 23.3% of patients of our series. It is now recognized that seriously ill COVID-19 patients are at higher risk of venous and arterial thromboembolic events (34-36) despite prophylactic or curative anticoagulation, compared to non-COVID-19 matched patients (35) . The higher percentage in our series compared to previous series (14, 37) could be explained by the fact that acute ischemic lesions could have been underestimated on CT scans in previous studies while our study only relied on MRI examinations. Coronaviruses use the angiotensin-converting enzyme 2 (ACE2) to enter cells, I n P r e s s a receptor which is expressed in the epithelia of the lung and small intestine, and also found in brain endothelial cells (5, 6) . It has been suggested that recruitment of immune cells by direct viral infection of the endothelium or immune-mediated, with a massive inflammatory cascade, may led to endothelial damage, resulting in stroke, thrombosis and hemorrhage (38) .

Perfusion abnormalities, which were related to seizure in 19.6% of patients in our series, were also described in a smaller series (8), although the mechanism was not discussed.

Many factors can account for the occurrence of seizures in COVID-19 patients: fever that lowers the seizure threshold, metabolic alterations, iatrogeny and potential changes related to encephalitis.

Lastly, extensive supratentorial white matter FLAIR hyperintensities (9, 11, 12) and unilateral abnormalities of the medial temporal lobe (9,26) described in previous studies were not seen in our series.

Our study has several limitations. First, clinical examination was limited in this context and CSF data were only available in about half of the participants. Second, most patients had no prior MRI scan. Finally, at this point, information on the outcome of patients was lacking. A cross-analysis correlating clinical, imaging, CSF and pathological data with patient outcome will be the scope of future studies.

This study provides a detailed description of neuroimaging findings in a series of COVID-19

patients. In addition to cerebrovascular thrombotic events, perfusion abnormalities, CLOCC, microhemorrhages involving the splenium of corpus callosum and ICU-related disorders, two MRI patterns were identified including white matter lesions with perivascular enhancement, that may reflect vasculitis lesions and/or inflammatory lesions with angiocentric I n P r e s s involvement, and basal ganglia abnormalities including substantia nigra involvement. Future imaging-pathological correlation studies will help determine whether there is a causal relationship between COVID-19 and brain MRI lesions. A longitudinal monitoring will also be necessary to assess the long-term neurological impact of COVID-19. I n P r e s s I n P r e s s I n P r e s s Abbreviation: SARS-CoV-2: severe acute respiratory syndrome coronavirus 2 I n P r e s s Main clinical characteristics of the whole clinical cohort (n=308) and in patients with de novo acute neurological symptoms, as described in Figure 1 . 

Neurologic Manifestations of Hospitalized Patients With Coronavirus Disease

Possible central nervous system infection by SARS coronavirus

Detection of SARS coronavirus RNA in the cerebrospinal fluid of a patient with severe acute respiratory syndrome

Severe neurologic syndrome associated with Middle East respiratory syndrome corona virus (MERS-CoV)

Human Coronaviruses and Other Respiratory Viruses: Underestimated Opportunistic Pathogens of the Central Nervous System? Viruses

Does SARS-Cov-2 invade the brain? Translational lessons from animal models

Neuroinvasive and neurotropic human respiratory coronaviruses: potential neurovirulent agents in humans

Neurologic Features in Severe SARS-CoV-2 Infection

Brain MRI Findings in Severe COVID-19: A Retrospective Observational Study

Imaging in Neurological Disease of Hospitalized COVID-19 Patients: An Italian Multicenter Retrospective Observational Study

Brain MRI Findings in Patients in the Intensive Care Unit with COVID-19 Infection

Neurovascular Complications in COVID-19 Infection: Case Series

Neurologic Involvement in COVID-19: Cause or Coincidence? A Neuroimaging Perspective

Atypical Deep Cerebral Vein Thrombosis with Hemorrhagic Venous Infarction in a Patient Positive for COVID-19

Imaging of acute disseminated encephalomyelitis

International Pediatric MS Study Group. Acute disseminated encephalomyelitis

Acute disseminated encephalomyelitis: an update

Punctate and curvilinear gadolinium enhancing lesions in the brain: a practical approach

CD8 encephalitis in HIV-infected patients receiving cART: a treatable entity

Chorea-ballismus with nonketotic hyperglycemia in primary diabetes mellitus

Review: neuropathology of acute phase encephalitis lethargica: a review of cases from the epidemic period

Encephalitis lethargica syndrome: 20 new cases and evidence of basal ganglia autoimmunity

Encephalitis lethargica: 100 years after the epidemic

Encephalitis Lethargica With Isolated Substantia Nigra Lesions Followed by a Second Encephalitis in a Child With Humoral Immunodeficiency

A first case of meningitis/encephalitis associated with SARS-Coronavirus-2

Sars-Cov-2: Underestimated damage to nervous system

Pathology and pathogenesis of severe acute respiratory syndrome

Guillain-Barré Syndrome associated with SARS-CoV-2 infection

Neuropathological Features of Covid-19

Neuropathology of COVID-19: a spectrum of vascular and acute disseminated encephalomyelitis (ADEM)-like pathology

Early postmortem brain MRI findings in COVID-19 non-survivors

Neurologic Complications of Extracorporeal Membrane Oxygenation: A Review

Prevalence of venous thromboembolism in patients with severe novel coronavirus pneumonia

High risk of thrombosis in patients in severe SARS-CoV-2 infection: a multicenter prospective cohort study

Central nervous system manifestations of COVID-19: A systematic review

Brain Imaging Use and Findings in COVID-19: A Single Academic Center Experience in the Epicenter of Disease in the United States

Endothelial cell infection and endotheliitis in COVID-19

The authors gratefully acknowledge the generous support of the Fédération Internationale de l'Automobile (FIA), the FIA Foundation and donors of Paris Brain Institute -ICM. We thank the Cohort COVID-19 Neurosciences (CoCo Neurosciences) study group for its participation to data collection for their contribution to the discussion and the team of Radiology technicians of the Pitié-Salpêtrière Hospital for the data acquisition.

This project was funded by the Paris Brain Institute (ICM). ICM was supported by an unrestricted donation from the Fédération Internationale d'Automobile (https://www.fia.com/fia), a non-profit making association. The research was also supported by funding from the program "Investissements d'avenir" ANR-10-IAIHU-06.

The characteristics of the microhemorrhages were consistent with cerebral amyloid angiopathy in one non-ICU patient. 3 ASL was available and interpretable in 46 participants including 22 non-ICU and 24 ICU patients. In two patients, perfusion abnormalities could be attributed both to seizure and ischemia. 4 Enhancement of the corpus callosum signal abnormality spot was seen in one patient; no contrast administration was performed in the two other cases They were potentially related to leukostasis in an ICU patient with chronic lymphocytic leukemia 5 Contrast material administration was performed in 42 patients (21 non-ICU and 21 ICU patients). Dural enhancement in two patients was likely due to lumbar puncture; an additional leptomeningeal enhancement was seen with one of these patients. 6 Including optic neuritis and vestibulocochlear neuritis. Coagulation tests findings in patients with 1 to 4 microhemorrhages (first two columns), in patients with ≥5 microhemorrhages (third and fourth columns) and in the patient with a cerebral deep cerebral vein thrombosis (last column). Results are expressed in mean ± SD, range 1 A T2 star or susceptibility-weighted sequence were missing in two patients (one ICU and one non-ICU), explaining the number of 71 (instead of 73).Abbreviations: ICU: intensive care unit