A review of neuroradiological abnormalities in patients with coronavirus disease 2019 (COVID-19).

BACKGROUND: Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has led to various neurological manifestations. There is an urgent need for a summary of neuroimaging findings to accelerate diagnosis and treatment plans. We reviewed prospective and retrospective studies to classify neurological abnormalities observed in patients with the SARS-CoV-2 infection.
METHODS: The relevant studies published in Scopus, PubMed and Clarivate Analytics databases were analysed. The search was performed for full-text articles published from 23 January 2020 to 23 February 2021.
RESULTS: In 23 studies the number of patients with SARS-CoV-2 infection was 20,850 and the number of patients with neurological manifestations was 1996 (9.5%). The total number of patients with neuroradiological abnormalities was 602 (2.8%). SARS-CoV-2 has led to various neuroimaging abnormalities which can be categorised by neuroanatomical localisation of lesions and their main probable underlying pathogenesis. Cranial nerve and spinal root abnormalities were cranial neuritis and polyradiculitis. Parenchymal abnormalities fell into four groups of: (a) thrombosis disorders, namely ischaemic stroke and sinus venous thrombosis; (b) endothelial dysfunction and damage disorders manifested as various types of intracranial haemorrhage and posterior reversible encephalopathy syndrome; (c) hypoxia/hypoperfusion disorders of leukoencephalopathy and watershed infarction; and (d) inflammatory disorders encompassing demyelinating disorders, encephalitis, vasculitis-like disorders, vasculopathy and cytotoxic lesions of the corpus callosum. Leptomeninges disorders included meningitis. Ischaemic stroke was the most frequent abnormality in these studies.
CONCLUSION: The review study suggests that an anatomical approach to the classification of heterogeneous neuroimaging findings in patients with SARS-CoV-2 and neurological manifestations would lend itself well for use by practitioners in diagnosis and treatment planning.

Introduction

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), as a novel coronavirus, was first identified in Wuhan, China. It caused an outbreak of severe pneumonia in China that rapidly spread to the world. A high percentage of patients with SARS-CoV-2 infection (36%) have presented with central nervous system (CNS) symptoms, including headache, encephalopathy, stroke, meningoencephalitis and seizures.

The olfactory bulb has been suggested as one of the probable routes of viral entry to the brain. Other proposed routes include haematogenous spread and passage through the blood–brain barrier (BBB). The SARS-CoV-2 virus binds to angiotensin-converting enzyme II (ACE-II) receptor stop as the cell membrane; these receptors exist abundantly on cerebral vascular endothelial and neuron cells.[2,3] In comparison with other viruses, such as herpes simplex virus, some enteroviruses and some arthropod-borne viruses that are highly neurovirulent and associated with neuron destruction, SARS-CoV-2 or related coronaviruses are not very neuroinvasive. Although some studies have confirmed the presence of SARS-CoV-2 ribonucleic acid (RNA) in the patients’ cerebrospinal fluid (CSF)[4-6] or have even found the virus particles in the involved neurons, the role of this virus in the pathogenesis of neurological manifestations is not yet clear.[8,9] There is no evidence in favour of CNS disorders due to direct viral invasion. On the contrary, the role of indirect mechanisms such as the neuroinflammatory and autoimmune processes has been evidenced in CNS disorders secondary to SARS-CoV-2. The other mechanisms by which SARS-CoV-2 might have contributed to the occurrence of neurological disorders include endothelial damage, thrombosis and secondary hypoxia. These mechanisms result in neuroradiological abnormalities whose diagnosis on neuroimages is not straightforward due to the novelty of the virus as well as the heterogeneity of presented lesions. It is important to determine the nature of these disorders immediately, because some neuroradiological abnormalities, such as lesions secondary to the inflammatory process and thrombotic events, are treatable on early therapeutic intervention. To fill the gap, we reviewed prospective and retrospective observational studies to determine CNS abnormalities on the neuroimages of patients with SARS-CoV-2 infection. Not only does this anatomical classification of neuroradiological abnormalities assist practitioners with early diagnosis and individualised treatment plans, but this classificatory description also benefits the bio-medical data scientist in developing diagnosis algorithms.

Methods

The checklist and flow diagram of the Preferred Reporting Items for Systemic review and Meta-Analysis (PRISMA) were used to perform a literature review. Relevant studies were searched in Scopus, PubMed and Clarivate Analytics (formerly known as ISI Web of Knowledge).

Search strategy

Using the title, abstract and full-text fields, the search was conducted by two of the authors (AH and FN). The databases were searched using the following search string:

TS = ((Covid-19 OR SARS-CoV-2 OR coronavirus OR Novel coronavirus) AND (Brain OR Cerebral OR Brainstem OR Spinal OR Olfactory Bulb OR Cortex OR White Matter OR Gray Matter OR Basal Ganglia OR Thalamus) AND (MRI OR CT OR MR Angiography OR MR Venography OR CT Venography OR CT Angiography OR Angiography)) AND Period = 23/1/2020–23/2/2021.

The search was performed on 23 February 2021. The search string retrieved the documents with the search terms in their titles, abstracts or keywords. The retrieved studies were separately reviewed by three authors (SM, BB and KM).

Eligibility criteria

Inclusion criteria

All published documents that reported abnormal findings related to SARS-CoV-2 infection in the brain and spinal cord magnetic resonance imaging (MRI) or computed tomography (CT) were selected. The selected retrospective and prospective observational studies met the following eligibility criteria: (a) diagnosis of SARS-CoV-2 infection on the basis of a positive reverse-transcription polymerase chain reaction (RT–PCR) assay on swabs or clinical and laboratory parameters (including chest radiology and inflammation markers) suggestive of coronavirus disease 2019 (COVID-19); (b) sample size of 10 patients or more with abnormal findings in the neuroimaging study; (c) reporting neuroimaging findings secondary to SARS-CoV-2 infection; (d) providing a detailed description of imaging findings (i.e. MRI and CT); and finally (e) suggesting presumptive neuroradiological diagnosis in the results or discussion sections for meningitis, stroke, sinus venous thrombosis (SVT), leukoencephalopathy, microhaemorrhage, intracranial haemorrhage, dissection and posterior reversible encephalopathy syndrome (PRES), temporal lobe encephalitis (TLE), acute necrotising encephalitis/encephalopathy (ANE), acute disseminated encephalomyelitis (ADEM), encephalitis with nigrostriatal pathway involvement, cytotoxic lesions of the corpus callosum (CLOCC), vasculitis-like disorders and vasculopathy.

Exclusion criteria

Studies were excluded if they were: (a) hypotheses, case reports, case series less than 10, editorials and letters; (b) papers reporting abnormal neuroimaging findings in Middle Eastern respiratory syndrome (MERS)-CoV and SARS-CoV patients; (c) documents without full texts; (d) articles published in languages other than English; (e) documents describing and analysing neurological manifestations of SARS-CoV-2 infection with normal neuroimaging findings; and (f) papers considering the old finding of neuroimaging studies as the main results and failing to distinguish them from the findings related to SARS-CoV-2.

Data extraction

Eligible studies were once more evaluated by five other authors for inclusion (RN, ARK, RE, AGN and MA). In case of any disagreement between the reviewers, the main author evaluated the eligibility of the articles and made the final decision (RN).

The data were manually extracted from the selected articles by the authors (SM, RE, ARK and AGN). They reviewed the full texts of the articles after the first screening process. In this phase, the desired data items were extracted including the first author, country, date, study characteristics, number of patients with neurological manifestations, mean and/or range of age, sex, imaging modality and interpretation, number of patients with imaging findings, neuroimaging findings with presumptive diagnosis in results or discussions, neuroimaging findings without diagnosis and the number of histopathological examinations.

Results

Study selection

Initially, 979 studies were retrieved from databases including Scopus, Clarivate Analytics and PubMed. After excluding duplicate records and non-original articles, 539 original articles were identified. After screening these articles based on their titles and abstracts, 407 articles were excluded and 31 records remained. These articles were reassessed through full-text evaluation and eight more articles were excluded. Finally, 23 articles were included in the qualitative synthesis. Figure 1 shows the flow diagram of the included studies according to the PRISMA for systematic reviews. The number of patients with SARS-CoV-2 infection was 20,850 and the number of patients with neurological manifestations was1996 (9.5%). The total number of patients with neuroradiological abnormalities was 602 (2.8%) and the number of abnormal imaging findings was 794. Some patients had two or more patterns of abnormality in the neuroimaging studies. The clinical characteristics of the included studies are summarised in Table 1. ,[14-35]

Figure 1.

The preferred reporting items for systematic reviews and meta-analysis (PRISMA) flow diagram.

Table 1.

The clinical characteristics of the included studies.

First author	Country	Date	Study design	No. of patients with neurological manifestation	Age (range)	Male	No. of patients with neuroimaging findings	Neuroimaging modality	Image analysis	Imaging abnormality with diagnosis*	Imaging abnormalities without diagnosis	Histopathology
Radmanesh 14	USA	April 5–April 25, 2020	Retrospective	22	53 (38–64)	9	22	MRI	2 Neuroradiologists	CLE: 10 MH: 7 AIS: 11
Nawabi 15	Germany, France, Switzerland	February 16–May 19, 2020	Retrospective	18	49.5 (39.5–62.8)	9	18	CT and MRI	NA	IPH: 6 SAH: 11 SDH: 1 IVH: 3 MH: 1
Kremer 16	French	March 16–April 9, 2020	Retrospective	68	66 (20–92)	43	36	MRI	3 Neuroradiologists	AIS: 11 LME: 5 TE: 2 ANE: 2 CLOCC: 1 ADEM: 1 Miscellaneous encephalitis: 2
Scullen 17	USA	April 22, 2020	Retrospective	27	59.8 (35–91)	14	27	MRI and CT	NA	ANE: 2 AIS: 4 Vasculopathy‡: 5 IPH: 3	Deep lobar hypoattenuation: 6Deep supratentorial hypodensity: 14
Helms 18	French	March 3–May 5, 2020	Prospective	118	62	89	28	MRI	NA	MH: 8 LME: 17 AIS: 3	FLAIR hypodensity: 4
Klironomos 19	Sweden	March 2–May 24, 2020	Retrospective	185	62	138	NA (CT: 53 MRI: 43)	CT and MRI	NA	IPH: 11 SAH: 5 SDH: 8 EDH: 2 MH: 29 CLE: 18 AIS: 15 Vasculopathy‡: 6 CLOCC: 1 LME: 3 OB involvement: 7 Root involvement: 2 CNA: 2	GP restriction on DW with enhancement: 1 White matter changes: 22
Hernández-Fernández 8	Italy	March 1–April 19, 2020	Retrospective	23	66.9	18	23	CT and MRI	1 Neuroradiologists	AIS: 17 IPH: 5 PRES: 1		6
Radmanesh 20	USA	March 1–31, 2020	Retrospective	242	68.7	150	24	CT and MRI	1 Neuroradiologist	AIS: 13 IPH: 11 CLE: 1
Kremer 21	French	March 23–April 27, 2020	Retrospective	37	61	30	37	MRI	3 Neuroradiologists	TLE: 16 MH: 9 CLOCC: 2 ANE: 2	Multifocal WM enhancing lesions with or without haemorrhage: 13 MCP hypodensity: 2
Chougar 22	French	March 23–May 7, 2020	Retrospective	73	58.5 (28–96)	48	43	MRI	2 Neuroradiologists	AIS: 17 SVT: 1 MH: 20 CLOCC: 3 PRES:2 CLE: 3 CPM: 3 LME: 2 Neuritis: 2 ENSPI: 4 Vasculitis-like: 4	Corticospinal hypodensity: 2
Pilloto 23	Italy	February 20–May 31, 2020	Retrospective	25	65.9 (50–84)	15	12	MRI	NA	ADEM: 1 ANE: 2 TLE: 2 Encephalitis with NS pattern†: 7
Paterson 24	England	9 April and 15 May, 2020	Retrospective	43			30	MRI and CT	NA	ADEM: 8 Myelitis: 1 TLE: 2 AIS: 8 MH:1 Root enhancement: 1 Dilated optic sheet in ICP: 1 CLE: 1 Encephalopathy with NS lesion†: 3
Yoon 25	USA	March 3–May 6, 2020	Retrospective	150	63.6 (22–96)	98	26	MRI and CT	2 Neuroradiologists	MH: 7 IPH: 2 SAH: 2 AIS: 14 CLE: 7
Sawlani 26	England	March 1–May 31, 2020	Retrospective	167	NA	NA	38	MRI and CT	2 Neuroradiologists	MH: 12 AIS: 5 MT: 3 ANE: 2 ADEM: 1 IPH: 2 SAH: 1 CLE: 1	Watershed hyperintensity: 4
Mahammedi 27	Italy	February 29–April 4, 2020	Retrospective	108	69	69	51	MRI and CT	4 Neuroradiologists	AIS: 36 ICH: 6 SVT: 2 MS plaque exacerbation: 2 Encephalopathy: 1 GBS: 2 MFS: 1 PRES: 1	T2/FLAIR hyperintensity: 12
Lin 28	USA	March 4–May 9, 2020	Retrospective	278 (imaging)	71.8	165	58	MRI and CT	2 Neuroradiologists	AIS: 31 IPH: 10 SDH: 3 SAH: 2 CNA: 6 PRES: 3 MH: 3
Büttner 29	Germany	March 15–April 24, 2020	Retrospective	34	59	10	13	MRI and CT	4 Neuroradiologists	MH: 7 SAH: 4 CLE: 4 ICH: 2 AIS: 2 Superficial haemosiderosis: 3
Conklin 30	USA	March 12–May 14, 2020	Prospective	16	47–78/on supp	11	16	MRI	2 Neuroradiologist	MB: 11		1
D’Amore 31	Italy	February 21–May 21, 2020	Prospective	27	NA	NA	15	MRI and CT	3 Neuroradiologists	AIS: 6 PRES: 1 SAH: 1 SDH: 2 IPH: 3 ICH (multicompartment): 1 Encephalitis: 1
Dixon 32	England	April 1–June 1, 2020	Retrospective	28	56	8	10	MRI	2 Radiologists	MH: 10
Greenway 33	USA	July 3–November 23, 2020	Retrospective	180 (imaging)	20–94	104	27	MRI and CT	–	Acute/subacute stroke: 13 ICH: 6 (IPH: 5, SDH: 2, SAH: 1) PRES: 1	Other: 7
Lersy 34	France	March 1–May 31, 2020	Retrospective	69	69 (64 − 79)	46	11	MRI	2 Neuroradiologists	Vasculopathy‡: 11 Concomitant with, AIS: 4 MH: 6 SAH: 3	Parenchymal hyperintensity: 2
Lersy 35	France	1 March–May 31, 2020	Retrospective	58	62 (55–70)	38	37	MRI	2 Neuroradiologists	LME: 23 CNA: 1 MH: 12 AIS: 9 SVT: 2 Acute demyelinating lesions: 2	Diffuse WM hyperintensity: 4 MCP hyperintensity: 2 Temporal hyperintensity: 1 GM hyperintensity: 1

*Some patients had two or more patterns of abnormality in neuroimaging studies.

†Considered as miscellaneous encephalitis/encephalopathy.

‡Concentric arterial wall enhancement on post-contrast MRI.

ADEM:- acute disseminated encephalomyelitis; AHEM: acute haemorrhagic encephalomyelitis; AIS: acute ischaemic stroke; ANE: acute necrotising encephalopathy; CLOCC: cytotoxic lesions of the corpus callosum; CLE: cerebral leukoencephalopathy; CAN: cranial nerve abnormality; CT: computed tomography; EDH: epidural haemorrhage; ENSPI: encephalitis with nigrostriatal pathway involvement; GBS: Guillain–Barré syndrome; GM: gray matter; GP: globus pallidus; ICH: intracranial haemorrhage; IPH: intraparenchymal haemorrhage; TLE: temporal lobe encephalitis; LME: leptomeningeal enhancement; MCP: middle cerebral peduncle; MFS: Miller Fisher syndrome; MH: microhaemorrhage; MS: multiple sclerosis; MRI: magnetic resonance imaging; PRES: posterior reversible encephalopathy syndrome; SAH: subarachnoid haemorrhage; SDH: subdural haemorrhage; SVT: sinus venous thrombosis; WM: white matter.

Neuroradiological disease classification

Comprehensive classification of neuroradiological abnormalities in the CNS is a helpful method for a rapid and timely diagnosis of these lesions. Following a two-way classification of HIV infections proposed by Price, we classified these abnormalities based on: (a) the neuroanatomical localisation of lesions; and (b) their main probable underlying pathogenesis. Firstly, it yielded three major neuroanatomical groups of cranial nerve and nerve root, parenchymal and leptomeninges layer. The extraction of the suggested neuro-mechanisms from the reviewed medical articles revealed the contribution of several pathophysiological mechanisms in the development of various CNS abnormalities by SARS-CoV-2 infection. Thus, in the second step, the CNS disorders of these three neuroanatomical groups were matched with the relevant mechanism groups including thrombosis, endothelial cell dysfunction and damage, hypoxia/hypoperfusion and inflammation (Figure 2). Figure 3 outlines a classification of SARS-CoV-2 infection diseases in the patients based on this two-way classification. The inflammatory mechanism has played a main role in the development of the diseases related to the cranial nerve and nerve root and leptomeninges groups, associated with neuritis, radiculitis and meningitis, respectively. Brain and spinal parenchymal abnormalities had a more complex pattern of occurrence as they were mediated by multiple mechanisms, but they were classifiable using the above-mentioned pathophysiological mechanisms (Figure 2).[8,14-35] Parenchymal abnormalities fell into four groups: (a) thrombosis disorders, i.e. ischaemic stroke and sinus venous thrombosis; (b) endothelial dysfunction and damage disorders encompassing various types of intracranial haemorrhage, diffuse microhaemorrhage, dissection and PRES; (c) hypoxia and hypoperfusion disorders presented as leukoencephalopathy and watershed ischaemic lesions; and (d) inflammatory disorders including ANE, ADEM, myelitis, encephalitis (TLE, encephalitis with nigrostriatal pathway involvement and miscellaneous encephalitis), vasculitis-like, vasculopathy and CLOCC.8,14--35

Figure 2.

Pathophysiological mechanisms of neuroradiological abnormalities of patients with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection. ANE: acute necrotising encephalopathy; CLOCC: cytotoxic lesions of the corpus callosum; CLE: cerebral leukoencephalopathy; HIP/ARS: hyperinflammatory phase/acute respiratory phase; MH: microhaemorrhage; PRES: posterior reversible encephalopathy syndrome; SVT: sinus venous thrombosis.

Figure 3.

Neuroradiological abnormalities of patients with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection. ADEM: acute disseminated encephalomyelitis; CLOCC: cytotoxic lesions of the corpus callosum; CLE: cerebral leukoencephalopathy; MH: microhaemorrhage; PRES: posterior reversible encephalopathy syndrome; SVT: sinus venous thrombosis.

The following subsections provide a more detailed description of the diseases according to this two-way classification.

Central nervous system

Cranial nerve and nerve root

The olfactory bulb involvement was seen in seven cases with SARS-CoV-2. MRI findings in these cases revealed olfactory bulbs and tract hyperintensity on fluid-attenuated inversion recovery (FLAIR) sequence with enhancement in two cases.

Cranial nerve involvement, defined as neuritis, was seen in 12 cases,[19,22,27,28,35] which had been secondary to Miller Fisher syndrome or Guillain–Barre syndrome in some studies,[27,35] as evidenced in Figure 4.

Figure 4.

(a)–(d) Facial and vestibular neuritis. (a) and (b) A woman in her early 60s with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and facial neuritis. Precontrast (a) and postcontrast (b) T1-weighted imaging (WI) brain magnetic resonance imaging (MRI) showed bilateral facial nerve enhancement (b, arrow). (c) and (d) A woman in her mid-40s with SARS-CoV-2 infection and right-sided vestibular neuritis. Precontrast (c) and postcontrast (d) T1-WI brain MRI showed enhancement of the right vestibular nerve (d, arrow). (e)–(h) Polyradiculitis in a woman in her mid-30s with SARS-CoV-2 infection, lumbar pain and leg paresthesia. (e) and (f) Precontrast (e) and postcontrast (f) sagittal lumbar T1-WI showed faint contrast enhancement of the cauda equina (f, arrow). (g) and (h) Precontrast (g) and postcontrast (h) axial T1-WI images showed faint enhancement of multiple nerve roots (h, arrow).

The neuroradiological abnormality found in the nerve root was of the inflammatory type, namely radiculitis. Three studies reported multiple nerve root enhancements in five patients with acute demyelinating polyneuropathy postcontrast MRI, which were suggestive of polyradiculitis , , (Figure 4).

Parenchymal abnormalities

Parenchymal abnormalities were derived from four main mechanisms including thrombosis, endothelial cell dysfunction and damage, hypoxia/hypoperfusion and inflammation.

Thrombosis disorders

Thrombotic events related to SARS-CoV-2 were ischaemic stroke and SVT. A total of 219 cases of ischaemic stroke were reported by 18 studies.[8,14,16-20,22,24-29,31,33-35] Various types of cerebral infarction were also seen: large vessel occlusion,[8,16-20,24,27,28,33] embolic stroke[8,22,27,28,30,33] and lacunar infarction.[19,27,29,33,34,37] Based on the Trial of ORG 10172 in Acute Stroke Treatment (TOAST) classification, Hernández-Fernández et al. reported that a large number of patients with stroke were in the undetermined origin group (52.9%), followed by cardiac emboli (23.5%), other determining causes (17.6%) and stroke with an atherothrombotic origin (6%). Anterior circulation was involved more than posterior circulation in SARS-CoV-2 patients with acute stroke, as reported in a study by Radmanesh et al. This finding was consistent with the other studies reporting stroke in SARS-CoV-2 patients. Cerebral stroke was also associated with several uncommon neuroradiological abnormalities in patients with SARS-CoV-2, including focal or diffuse medium and large-sized cerebral artery narrowing in angiography[8,17] and vessel wall enhancement on post-contrast MRI.[8,19,34] The inflammatory aspects of these findings are explained below in the relevant section.

The histopathological examination of extracted arterial clots in four patients with ischaemic stroke revealed that the thrombus tissue mainly comprised fibrin and red blood cells with a variable number of platelets.

Another thrombotic event in patients with SARS-CoV-2 infection was SVT. Seven cases of SVT were reported by four studies.[20,22,27,35]

Endothelial dysfunction/damage-related disorders

SARS-CoV-2 infection by endothelial cell dysfunction and damage leads to various neuroradiological abnormalities encompassing ICH, PRES and dissection. Fifty-eight cases of intraparenchymal haemorrhage (IPH) were reported in 10 studies.[8,15,17,19,20,25,26,28,31,33] A large number of IPH cases were lobar, whereas the deep locations and typical sites of IPH secondary to hypertension were not common. Of note, the location of IPH was similar to the location of haemorrhage secondary to PRES. In the study by Hernández-Fernández et al., two patients demonstrated PRES-like white matter involvement. The most striking findings in the histopathological examination of two patients with IPH were structural changes in the vessel wall, including the small arterioles, capillaries and venules, associated with the disappearance of endothelial cells and local inflammation. A non-significant inflammation process in the vessel wall was reported, and arteriolosclerosis change and cerebral amyloid angiopathy were not also present in the histopathological examination. The cerebral vascular studies of some cases did not show any primary vascular malformation,[8,19] and cortical arterioles were reported as the probable main origin of the haematoma in the pathological studies.

Extra-axial haemorrhage was reported in 51 patients with SARS-CoV-2 by nine studies,[15,19,25-29,31,34] including 30 cases with subarachnoid haemorrhage (SAH),[15,19,25,26,28,29,31,33,34] 16 patients with subdural haemorrhage (SDH),[15,19,28,31,33] two patients with epidural haemorrhage (EDH) and three patients with intraventricular haemorrhage (IVH). No aneurysm or other vascular malformation was detected in patients with SAH.

The pathophysiology of intracranial haemorrhage was not clear, but several factors such as increased blood pressure and the prophylaxis and treatment of thrombosis have been proposed. The tropism of virus to the endothelial cell layer of cerebral vessels by their ACE-II receptors may play a role in intracranial haemorrhage. A pathological study has detected virus particles in the endothelial cells and the concomitant inflammatory reaction associated with endothelial cell destruction.39 ACE-II receptors have a crucial function in the regulation of cerebral blood flow via the sympathoadrenal system and vascular autoregulation; thus virus-induced receptor dysfunction might have led to the loss of vascular autoregulation, episodic blood pressure elevation and eventually arterial wall rupture.

Another pathological finding related to endothelial damage in patients with SARS-CoV-2 infection was arterial dissection in two cases reported by Hernández-Fernández et al. One case was a 48-old-year man with SARS-CoV-2 infection and ischaemic stroke whose CT angiography showed dissection at the origin of the right internal carotid artery (ICA), with an intimal flap. In extracted thrombus tissues, apoptotic cells, endothelial cells and atypical lymphocytes were seen. Viable endothelial cells detached from the arterial wall and migrated towards the arterial lumen were noted. However, in another study by Hernández-Fernández et al., detached endothelial cells were not seen in the histopathological examination of thrombus materials in non-SARS-CoV-2 patients with stroke, and even some of these specimens had inflammatory reaction and infectious agents in thrombus without detached endothelial cells. The risk of spontaneous arterial dissection coupled with recent infection and endothelial injury has been shown by a previous study. Overall, these findings suggest the potential threat of SARS-CoV-2 virus to endothelial cells.

Nine cases of PRES were reported in six studies,[8,22,27,28,31,33] of which four cases were haemorrhagic,[8,42,43] as illustrated in Figure 5. Traditionally, PRES is known as a complication of acute episode of blood pressure elevation and the consequently impaired autoregulation of cerebral blood flow. However, other mechanisms such as infections, inflammation, or nanotoxicity have been proposed as contributing to the development of PRES. In some patients with SARS-CoV-2 infection, PRES occurs in the absence of severe hypertension, which supports the involvement of other factors such as systemic inflammation. Another cause of PRES might be the endothelial dysfunction secondary to the activation of the immune system in severe SARS-CoV-2 infection, leading to cerebral oedema by increasing the permeability of the BBB.

Figure 5.

Posterior reversible encephalopathy syndrome in a 66-year-old man with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and seizure. Fluid-attenuated inversion recovery (FLAIR) sequence showed cortical and subcortical hyperintensity in the occipital lobe extending to parietal and frontal lobes (a) and (d, arrow), without restriction on diffusion-weighted imaging (DWI) (b) and (e). The follow-up magnetic resonance imaging (MRI) indicated the complete disappearance of the lesions (c) and (f).

SARS-CoV-2 can indirectly elevate blood pressure following ACE-II receptor dysfunction, as mentioned above.

Diffuse microhaemorrhage was reported in 143 patients with severe to critical acute respiratory phase (ARS) in 15 studies , , , , ,[24-26],[28-30], , , (Figure 6). Diffuse microhaemorrhage has a predilection for the corpus callosum, juxtacortical and subcortical white matter of both cerebral hemispheres and in the cerebellum and brainstem. CSF sampling and RT–PCR were obtained in two patients who were negative for SARS-CoV-2 infection and demonstrated normal cytology.[14,20] Several hypotheses have been offered to explain microhaemorrhage formation in these patients, including virus-induced endothelial damage, consumption coagulopathy and hypoxia.[46-48] Endothelial damage may cause microhaemorrhage following the attachment of the viral particles to ACE-II receptors that are widely expressed in endothelial cells. Endothelial layer damage leads in turn to the disintegration of the cerebral capillary wall and subsequent microhaemorrhage formation.

Figure 6.

Leukoencephalopathy with diffuse microhaemorrhage in an elderly man with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and delay in awakening after sedation withdrawal. Brain magnetic resonance imaging (MRI) shows diffuse leukoencephalopathy with the involvement of subcortical on fluid-attenuated inversion recovery (FLAIR) (a) and (b) and diffuse microhaemorrhage with a predilection for subcortical U-fibres and the corpus callosum on susceptibility weighted imaging (SWI) (c).

Consumption coagulopathy is the other mechanism proposed to explain microhaemorrhage in patients with SARS-CoV-2 infection. The evidence for this suggestion comes from the increased levels of D-dimer and fibrinogen. It may be associated with microthrombosis in small medullary veins and blood stasis, leading to cerebral microbleeding.[46,47] The role of thrombosis in microhaemorrhage has been evidenced in the microthrombosis of cerebral vessels in the neuroimaging study and pulmonary vessels in the autopsy examination of patients with severe SARS-CoV-2 infection. Hypoxia has also been suggested as another factor contributing to microhaemorrage. Similar patterns of diffuse microhaemorrhage have been seen in the victims of carbon monoxide poisoning, drug overdose, cardiopulmonary arrest and high-altitude exposure. Hypoxaemia results in the disruption of the BBB and the extravasation of erythrocytes through endothelial cell necrosis, eventually leading to cerebral microbleeds. The pathophysiology of microhaemorrhage is not yet clear; however, it seems endothelial damage following different mechanisms plays a pivotal role in microhaemorrhage.

Hypoxia/hypoperfusion-related disorders

Cerebral leukoencephalopathy and watershed infarction were identified as neuroradiological abnormalities related to hypoxia/hypoperfusion. Leukoencephalopathy was reported in 45 cases with SARS-CoV-2 by eight studies,[14,19,20,22,24-26,29] as illustrated in Figure 6. It was mainly reported in patients with a severe to critical ARS and was detected after sedation withdrawal. MRI findings were T2 hyperintensities that were bilaterally symmetric and presented with confluent hyperintensity of subcortical and deep white matter on T2-WI, with mild restricted diffusion. Abnormalities involved the posterior area more than the anterior. The deep gray nuclei were spared. Post-contrast T1-WI showed no enhancement in one patient. The brainstem and cerebellum were involved to a lesser degree; however, mild involvements of middle cerebellar peduncles and medial cerebellar hemispheres were reported in some cases.

Watershed infarction, as a consequence of hypoperfusion, has also been described in patients with SARS-CoV-2 infection.[16,27,50]

Inflammation-related disorders

SARS-CoV-2 infection has resulted in various types of inflammatory lesions including demyelinating disorder, encephalitis, CLOCC, vasculitis-like disorders and vasculopathy.

Twelve cases of ADEM were reported in four studies[16,23,24,26] (Figure 7), of which four cases were of the haemorrhagic type, specifically acute haemorrhagic encephalomyelitis (AHEM). Paterson et al. reported a 47-year-old woman with SARS-CoV-2 infection and AHEM. Although she received a high dose of intravenous steroids (methylprednisolone), she developed early signs of brain herniation and underwent emergent right hemicraniectomy. The histopathological examination demonstrated an inflammatory demyelinating process without a vasculitis pattern, supporting the diagnosis of hyperacute ADEM. The brain tissue was, however, negative for SARS-CoV-2 RT–PCR. The ADEM case in the Paterson study had myelitis in spinal MRI with normal brain MRI.

Figure 7.

Acute haemorrhagic encephalomyelitis in a 52-year-old man with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection. Initial T2 weighted imaging (WI) showed confluent multifocal hyperintense lesions in the white matter, corpus callosum, internal and external capsulesonT2-WI (a) and (b), with progression in the follow-up study, ending in diffusely involved white matter (d) and (e). Susceptibility weighted imaging (SWI) showed microhaemorrhages changes (c) and (f, arrow) and prominent medullary veins (f, short arrows).

Various types of encephalitis were reported in patients with SARS-CoV-2, including TLE (Figure 8), ANE (Figure 9), encephalitis with nigrostriatal pathway involvement (Figure 10) and miscellaneous encephalitis. Twenty-six cases of TLE with abnormal MRI findings were reported in four studies.[16,21,23,24] Ten cases of ANE were reported by five articles.[16,17,21,23,26] Of these patients, a 43-year-old female patient with SARS-CoV-2 infection and multi-organ failure developed neurological manifestations after sedation withdrawal and ventilator weaning. Brain MRI showed hyperintensity in bilateral mesial temporal structures, lenticular nuclei, cruscerebri and centrum semiovale on FLAIR. DWI sequence revealed a restriction of the same areas accompanying the splenium, body and genu of the corpus callosum. Haemorrhagic change was also seen in the left cerebral peduncle and the bilateral basal ganglia on the SWI sequences; these findings favoured the diagnosis of ANE. Chougar et al. reported four intensive care unit (ICU) patients with COVID-19 with abnormal signal changes in the nigrostriatal pathway and the presumptive diagnosis of encephalitis with nigrostriatal pathway involvement. They presented with motor or extrapyramidal symptoms after the withdrawal of sedation (Figure 10). All cases tested negative for SARS-CoV-2 RNA in their CSF analysis. The neuroimaging findings of the patients with encephalitis with nigrostriatal pathway involvement revealed: (a) signal and diffusion abnormalities, with variable contrast enhancement, affecting the substantia nigra, the globus pallidus and the striatonigral pathway; (b) hyperintensity of the substantia nigra on FLAIR and diffusion restriction on DWI with corresponding decreased apparent diffusion coefficient (ADC) in one case; (c) increased signal of the globus pallidus on the T1-WI MPRAGE sequence, without enhancement on the post-contrast T1-WI in another case; (d) diffusion restriction on DWI with corresponding decreased ADC in the globus pallidus, with enhancement on post-contrast T1-WI in one case. The authors proposed that this pattern of involvement had some similarities to that seen in patients with influenza suffering from encephalitis lethargica during the 1918 influenza pandemic. The patients with encephalitis lethargica presented with pharyngitis, sleepiness, ocular motility and movement disorders in the acute phase, which was fatal in a percentage of the patients. Some patients had delayed Parkinsonism and neuropsychiatric signs. Brain MRI showed bilateral oedematous changes in the thalamus, basal ganglia and midbrain, with variable contrast enhancements. Infectious and environmental causes were proposed as the possible aetiologies of encephalitis lethargica. Histopathological examination failed to show the presence of influenza RNA in the cerebral tissue. On the other hand, the presence of oligoclonal bands (OCBs) in the CSF and the vital role of corticosteroids in the treatment of patients suggested that encephalitis lethargica might be an immune-mediated disease. Fourteen cases of miscellaneous encephalitis were reported in five studies.[16,23,24,27,31] Neuroimaging findings in this group were unspecified and had different patterns compared to the usual findings mentioned above for ADEM, TLE, ANE and other encephalitides.

Figure 8.

Temporal lobe encephalitis in a 58-year-old man with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection. Fluid-attenuated inversion recovery (FLAIR) sequence showed hyperintensities (arrow) in the left medial temporal lobe.

Figure 9.

Acute necrotising encephalitis in a 59-year-old woman with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection. T2-weighted image (WI)sequence showed extensive hyper signal lesions of the bilateral medial temporal lobes, basal ganglia, thalami and pons (a), and a restriction on diffusion-weighted imaging (DWI) (c) with punctate haemorrhage change on susceptibility-weighted imaging (SWI) (b) and enhancement on post-contrast T1-WI (d, arrow).

Figure 10.

Encephalitis with nigrostriatal pathway involvement in a 42-year-old man with chronic lymphocytic leukaemia and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection. Brain magnetic resonance imaging (MRI) shows a diffusion restriction of the substantia nigra on diffusion-weighted imaging (DWI) (a, arrows) with corresponding decreased signal on apparent diffusion coefficient (ADC) (b, arrows) and hyperintensity on fluid-attenuated inversion recovery (FLAIR) (c, arrows).

Four studies reported seven cases with CLOCC in patients with SARS-CoV-2 infection , , , (Figure 11). The MRI findings included a non-enhancing hyperintense lesion within the splenium of the corpus callosum on FLAIR and DWI sequences with decreased signals on ADC sequence. CLOCC as a clinicoradiological syndrome known by a transient mild encephalopathy and a reversible oedematous lesion of the corpus callosum was mainly observed in the splenium on MRI and was associated with a transient encephalopathy. CLOCCs occur by numerous aetiologies, but the infection is identified as the most frequent cause of this syndrome. They occurred following a cytokine dysregulation and elevation inducing glutamate release in the extracellular space, resulting in neurons and microglia cell dysfunction with water influx into the intracellular space and an eventual cytotoxic oedema.

Figure 11.

Cytotoxic lesion of the corpus callosum (CLOCC) in a woman in her late 40s with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). It showed a hypersignal lesion in the posterior portion of the corpus callosum on T2-weighted image (WI) and fluid-attenuated inversion recovery (FLAIR) (c) and (d) with restriction on diffusion-weighted imaging (DWI) (a) and corresponding decreased signal on apparent diffusion coefficient (ADC) (b).

Chougar et al. reported a cerebral vasculitis pattern in the brain MRI of patients with SARS-CoV-2 infection. The authors reported four cases with late awakening after sedation withdrawal in the ICU. Their brain MRIs showed bilateral non-confluent multifocal hypersignal lesions in deep and periventricular white matter, with variable signal changes on DWI and ADC associated with an enhancement of the perivascular spaces. These lesions had a vacuolated necrotic appearance in one patient observable in Figure 12. The vasculitis-like lesions have been reported on cerebral MRI,[22,51] and histopathological studies of patients with SARS-CoV-2 infection. The pattern of enhancement on brain MRI was suggestive of a vasculitis phenomenon with an angiocentric character rather than a demyelinating disorder. The pattern of perivascular enhancement has been shown in vasculitis disorders with angiocentric infiltrates such as neurolupus and neurosarcoidosis. Endothelitis induced by the direct virus invasion or the host inflammatory response has been reported by Varga et al. in patients with SARS-CoV-2 infection and can lead to cerebral ischaemia and inflammation. Reichard et al. reported neuropathological findings of a decedent with SARS-CoV-2 infection that included combined features of both a perivascular inflammation and a demyelinating disease. Similar cases of vasculitis-like lesions were reported by Hanafi et al. and Brun et al.

Figure 12.

Cerebral vasculitis-like lesion in a 50-year-old man with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection. Brain magnetic resonance imaging (MRI) shows multiple punctiform and oedematous lesions in the periventricular white matter which are hypersignal on fluid-attenuated inversion recovery (FLAIR) (a) and (e, arrow), a diffusion restriction on the diffusion-weighted imaging (DWI) (b) and (f, arrow) without decreased apparent diffusion coefficient (ADC) (c) and (g, arrow), and a perivascular enhancement on the post-contrast T1-weighted image (WI) (d) and (h, arrow).

As mentioned above, vasculopathy was reported in some patients with SARS-CoV-2 infection. It has been described as vessel narrowing on angiography (Figure 13) or vessel wall enhancement on post-contrast MRI , , , (Figure 14). Scullen et al. reported a 53-year-old male patient with simultaneous manifestation of SARS-CoV-2 infection, massive ischaemic stroke, vasculopathy and critical ARS. Magnetic resonance angiography (MRA) showed focal stenosis of the supraclinoidal segments of the left ICA with patency of the middle cerebral artery (MCA) M2 trunk. Hernández-Fernández et al. and Klironomos et al. reported six cases of ischaemic stroke with a diffuse arterial narrowing on the neuroimaging studies.[8,19] Lersy et al. reported a 69-year-old man with delay in awakening after sedative withdrawal. His post-contrast brain MRI showed vessel wall enhancement of the basilar artery, the left MCA and the right posterior cerebral artery (PCA), as well as narrowing of both PCAs on brain MRA. Cerebral vasculopathy with small and large vessel involvement has previously been observed in patients with HIV-1, HIV-2, and varicella zoster virus (VZV) infection. Vasculopathy associated with HIV and VZV may lead to both lacunar and large ischaemic strokes without any evidence of large arterial occlusion on non-invasive imaging. A pattern of thin circumferential contrast enhancement of the vessel wall was reported in patients with VZV vasculopathy. Histological examination of the arterial tissue in patients with VZV vasculopathy showed thickening of the intima layer secondary to the appearance of myofibroblasts expressing alpha smooth muscle actin that resulted in luminal narrowing/occlusion and ischaemic stroke. The VZV-infected arteries contain abundant neutrophils, as well as CD4+ and CD8+ T cells and CD 68+ macrophages in the adventitia, with low attendance of B cells expressing CD20. Although the nature of neuroimaging abnormalities in SARS-CoV-2 patients remains unknown due to lack of histological examinations, the prevalence of inflammatory markers, as well as response to the high-dose steroid therapy, strengthens the role of inflammatory processes in the occurrence of abnormalities. It should be noted that vasculopathy and vasculitis might be confused with each other and have been used in literature reviews interchangeably,[34,57] but they seem to be two different entities, at least in neuroimaging studies. Vasculopathy is characterised by arterial narrowing in vascular imaging studies, vessel wall enhancement on post-contrast MRI and typical cerebral infarction. Vasculitis is, on the contrary, characterised by parenchymal lesions with punctiform enhancements on post-contrast MRI and without vascular narrowing in vascular imaging studies, which is suggestive of small vessel vasculitis.[51,54] These two neuroradiological abnormalities might be the extreme presentations of a common pathology continuum, in which vasculopathy has resulted from macro-vascular involvement and vasculitis from micro-vascular involvement; however, histological evaluation substantiating the nature of these pathologies is needed.

Figure 13.

Vasculopathy (arterial narrowing) and stroke in a 54-year-old male patient with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection. Non-contrast computed tomography (CT) of the head showed initial temporal haemorrhage (a, arrow) with surrounding parenchyma oedema. Magnetic resonance angiography (MRA) showed focal stenosis of the supraclinoid internal carotid artery (b, arrow) with a patent lumen and flow distal to stenosis (b, double arrows). Diffusion-weighted imaging (DWI) revealed a restriction in the right middle cerebral artery (MCA) territory, suggestive of acute cerebral infarction (c) and (d, arrow).

Figure 14.

Vasculopathy (arterial narrowing and vessel wall enhancement) in a 69-year-old male patient with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and delay in awakening after sedative withdrawal. (a)–(f) Precontrast T1-weighted image (WI) was normal, but post-contrast T1-WI showed vessels wall enhancement of the basilar artery (b) and (c), the left middle cerebral artery (d) and the right posterior cerebral artery (PCA) (e), as well as narrowing of both PCAs on brain magnetic resonance angiography (MRA) (f).

Leptomeninges

Leptomeningeal enhancement, known as meningitis, has been reported in 50 cases with SARS-CoV-2 infection by five studies.[18,19,22,23,35] Helms et al. and Lersy et al. used both post-contrast three-dimensional (3D) FLAIR and T1-WI in SARS-CoV-2 patients with neurological symptoms, and reported patients with leptomeningeal enhancement more frequently compared with other studies that used post-gadolinium T1-WI alone. Post-contrast enhancement appears to be a common imaging finding in patients with concomitant SARS-CoV-2 pneumonia and neurological symptoms. This abnormality may not be visualised well on post-gadolinium T1-WI; therefore, adding a post-contrast 3D FLAIR image may result in the more sensitive detection of the leptomeningeal signal changes and meningeal involvement in COVID-19 patients.[35,53] In SARS-CoV-2 patients, leptomeningeal enhancement can persist, even after clinical improvement. Post-contrast MRI has shown meningeal enhancement in patients with multiple sclerosis, which has an association with OCB in the CSF. OCB has been reported in CSF samples of SARS-CoV-2 patients with leptomeningeal enhancement. Klironomos et al. reported one case with a meningeal post-contrast enhancement, whose follow-up MRI showed the progression of the leptomeningeal enhancement despite the clinical improvement.

Neuroradiological abnormalities without a presumptive diagnosis

Some of the neuroradiological abnormalities related to SARS-CoV-2 infection did not lead to any clinicoradiological diagnosis. Ninety-four cases with these findings were reported by nine studies.[9,18,19,21,22,26,27,34,35] The findings reported by the selected studies included diffuse white hyperintensity, gray matter hyperintensity, temporal hypersignal lesion, FLAIR hyperintensity,[18,27] middle cerebellar peduncle hyperintensity,[21,35] parenchymal hyperintensity, globus pallidus restriction on DWI with enhancement, white matter changes, multifocal haemorrhagic or non-haemorrhagic white matter lesions with variable enhancements, corticospinal hyperintensity and watershed hyperintensity. Some cases had a deep lobar hypoattenuation or deep supratentorial hypodensity on brain CT. It seems that major numbers of neuroradiological abnormalities without a presumptive diagnosis were secondary to hypoxia or the inflammatory process.

Frequency of neuroradiological diseases

The most frequent neuroradiological diseases were ischaemic stroke, followed by diffuse microhaemorrhage, encephalitis, intraparenchymal haemorrhage, extraxial haemorrhage, meningitis, leukoencephalopathy, demyelinating disorders, PRES and vasculitis-like disorders, respectively. However, a large number of findings fell in the category of imaging abnormalities without diagnosis; that is, other lesions. The frequency of neuroradiological abnormalities is shown in Figure 15.

Figure 15.

Frequency of neuroradiological abnormalities in 602 patients with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection.

Discussion

SARS-CoV-2 infection has led to various neuroimaging abnormalities which can be classified firstly based on the neuroanatomical localisation of lesions and then the main probable underlying pathogeneses. Cranial nerve and nerve root abnormalities were olfactory bulb oedema, cranial neuritis and polyradiculitis. Parenchymal abnormalities can be further grouped as: (a) thrombosis disorders that were ischaemic stroke and sinus venous thrombosis; (b) endothelial dysfunction and damage disorders subdivided into various types of intracranial haemorrhage, diffuse microhaemorrhage, dissection and PRES; (c) hypoxia/hypoperfusion disorders encompassing leukoencephalopathy and watershed infarction; and (d) inflammatory disorders manifested as ANE, ADEM, myelitis, encephalitis (TLE, encephalitis with nigrostriatal pathway involvement and miscellaneous encephalitis), vasculitis-like disorders, vasculopathy and CLOCC. Leptomeninges disorders were manifested as meningitis. Ischaemic stroke was the most frequent abnormality in these studies. The proposed classification has the potential to accelerate diagnosis and treatment plans.

A close inspection of the studies revealed that the complications originated from different mechanisms, as mentioned above. Although we use these main mechanisms to propose a classification of SARS-CoV-2 neuroradiological disorders, it is not a rule and the neuropathophysiology of SARS-COV-2 disease is more complex, and some of these disorders such as stroke, microhaemorrhage or PRES are multifactorial.[8,14,19,21,22,24]

Therapeutic interventions proposed for the parainfectious inflammatory disorders were corticosteroids, intravenous immunoglobulin and plasma exchange, although therapeutic responses have been variable and dependent on disease severity and the time of intervention. Treatments of other non-inflammatory disorders were limited to supportive care; however, in these disorders, the early management of the hyperinflammatory phase of SARS-CoV-2 including acute respiratory syndrome can alleviate neurological complications. Antithrombotic drugs may also be useful in thrombotic disorders.[8,23,24]

Limitations

This literature review has some limitations. Concerning the short time from the SARS-CoV-2 pandemic, it is not possible to perform a comprehensive study to investigate the nature and the relationship of the neurological manifestations and radiological and histopathological findings. A causation association cannot be made based on the limited available evidence and data. Moreover, many diagnoses are mainly based on assumptions. Also, a limited number of comprehensive studies have been performed to evaluate neuroimaging and histopathological findings, which makes it difficult to diagnose neurological symptoms.

Conclusions

This review suggests an approach to various neuroimaging findings and CNS disorders in SARS-CoV-2 infection based on anatomical regions and complemented by their underlying pathogenesis. This two-way classification can be used to facilitate diagnosis and management planning.

Highlights

Heterogeneous neuroradiological abnormalities resulting from the COVID-19 infection are classifiable based on neuroanatomical localisations and pathogenesis.

Ischaemic stroke was the most common neuroradiological abnormality in studies.

Additional abnormalities included sinus venous thrombosis, intracranial haemorrhage, dissection, PRES, leukoencephalopathy, microhaemorrhages, ANE, acute myelitis, ADEM, encephalitis, vasculitis-like disorders and CLOCC.

Click here for additional data file.

Supplemental material, sj-pdf-1-neu-10.1177_19714009211029177 for A review of neuroradiological abnormalities in patients with coronavirus disease 2019 (COVID-19) by Bahar Bahranifard, Somayeh Mehdizadeh, Ali Hamidi, Alireza Khosravi, Ramin Emami, Kamran Mirzaei, Reza Nemati, Fatemeh Nemati, Majid Assadi and Ali Gholamrezanezhad in The Neuroradiology Journal