Acute and chronic neurological disorders in COVID-19: potential mechanisms of disease.

Coronavirus disease 2019 (COVID-19) is a global pandemic caused by SARS-CoV-2 infection and is associated with both acute and chronic disorders affecting the nervous system. Acute neurological disorders affecting patients with COVID-19 range widely from anosmia, stroke, encephalopathy/encephalitis, and seizures to Guillain-Barré syndrome. Chronic neurological sequelae are less well defined although exercise intolerance, dysautonomia, pain, as well as neurocognitive and psychiatric dysfunctions are commonly reported. Molecular analyses of CSF and neuropathological studies highlight both vascular and immunologic perturbations. Low levels of viral RNA have been detected in the brains of few acutely ill individuals. Potential pathogenic mechanisms in the acute phase include coagulopathies with associated cerebral hypoxic-ischaemic injury, blood-brain barrier abnormalities with endotheliopathy and possibly viral neuroinvasion accompanied by neuro-immune responses. Established diagnostic tools are limited by a lack of clearly defined COVID-19 specific neurological syndromes. Future interventions will require delineation of specific neurological syndromes, diagnostic algorithm development and uncovering the underlying disease mechanisms that will guide effective therapies.

Introduction

Since its discovery in Wuhan, China in late 2019, coronavirus disease 2019 (COVID-19), the illness caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has resulted in over 3 million deaths (https://covid19.who.int/) and placed unprecedented pressure on social, economic and health care systems worldwide. Many survivors of the acute infection experience persistent and incapacitating neurological symptoms, which can have socio-economic and personal consequences. It is thus imperative that there is a thorough understanding of the evolving clinical syndromes and the underlying pathophysiological mechanisms, enabling rational therapeutic interventions to be expeditiously deployed.

Retrospective cohort studies from around the world report neurological signs and symptoms such as headache, altered mental status, seizures and stroke, in over a third of patients during the acute phase of the illness, positioning SARS-CoV-2 as an emerging neuro-pathogen. A similar proportion of infected individuals develop a post-infectious viral syndrome with diverse neuropsychiatric manifestations. Viral infections cause neurological impairments through multiple mechanisms, including direct infection of neurons, glia or endothelial cells within the nervous system resulting in acute cell death, as observed in herpes simplex virus type-1 (HSV-1) encephalitis. Alternatively, viruses, e.g. the human immunodeficiency virus type-1 (HIV-1), can persist in cellular reservoirs within the central (CNS) and perhaps peripheral (PNS) nervous system resulting in chronic inflammation and insidious progressive neurological damage. Among non-neurotropic viruses such as influenza and other respiratory viruses, systemic infection is associated with inflammation, metabolic and hormonal derangements, with vascular injury resulting in neurological disease. The host immune responses triggered during or following viral infections can also result in autoimmune damage of neural tissues, as observed in the PNS [e.g. Guillain–Barré syndrome (GBS)] and in the CNS [e.g. acute disseminated encephalomyelitis (ADEM) or acute transverse myelitis (ATM)]. Each of these mechanisms are implicated in SARS-CoV-2 infection and are addressed next. Of note, a multisystem inflammatory syndrome in children (MIS-C) has been described in paediatric cohorts with COVID-19; several cohort studies of children infected with SARS-CoV-2 report neurological disorders resembling those observed in adults including headache, encephalopathy, demyelinating disorders and stroke. This review provides an update on neurological manifestations of COVID-19 that concentrates on adults while also examining contemporary evidence for the neuropathogenic mechanisms implicated in SARS-CoV-2 infection (Fig. 1) and their relationship to current and potential therapies (Table 1).

Figure 1

Potential mechanisms of acute neurological disease in COVID-19. (A) Multiple pathogenic processes result in injury to the brain during COVID-19 including vascular abnormalities resulting in thromboembolism, microhaemorrhage and endotheliopathy with associated antiphospholipid antibodies (αPL) and disruption of the blood–brain barrier (BBB) leading to bioenergy failure, autoantibodies (e.g. αGQ1b, α-NMDA-R, α-CASPR2 and αLGI2) that target a range of neural antigens, and neuroinvasion with infection of neurons and astrocytes via ACE2 as well as associated systemic inflammation and innate neuroimmune responses (cytokine, chemokine, protease and reactive oxygen species production and release by microglia and astrocytes). Therapeutic interventions that have been reported or proposed are indicated with an asterisk. (B) In the PNS and spinal cord, GBS associated with anti-glycan antibodies (αGL), T-cell mediated transverse myelitis, as well as myositis have been reported in patients with COVID-19 that may be responsive to different therapies. PLEX = plasma exchange.

Table 1

Proposed neuropathogenic mechanisms in SARS-CoV-2 infection

Acute neurological syndromes	Proposed mechanisms	References	Proposed therapies
Anosmia/ageusia	Direct infection of olfactory bulb Inflammation of olfactory tract	Meinhardt et al.,14 Lu et al.15	None
Stroke	Hypercoagulability/endothelial damage	Hernández-Fernández et al.,16 Goshua et al.,17 Yaghi et al.18	Prophylactic anticoagulation is currently under investigation; no clear guidelines to date   Successful treatment with thrombolysis and mechanical thrombectomy reported
Encephalitis	Viral neuro-invasion	Nampoothiri et al.,19 Meinhardt et al.14	Favourable responses to systemic corticosteroids, tocilizumab, and PLEX are observed in a subset of cases
	Disrupted blood–brain barrier	Alexopoulos et al.20
	Autoimmunity	Cao et al.,21 Guilmot et al.,22 Pilotto et al.23,24
Encephalopathy	Metabolic dysfunction   Hypoxia/ischaemia   Cerebral microthrombi   Cytokine storm (systemic)	Bryce et al.,25 Antony and Haneef,26 Lee et al.,27 Lin et al.28	Generally supportive, reported success with tocilizumab in case reports
Peripheral neuropathy	Critical illness neuropathy	Cabañes-Martínez et al.29	Supportive
	Molecular mimicry (GBS and variants)	Dalakas,30 Temme et al.31	Standard therapy: IVIg, PLEX
Myositis	Bioenergetic dysfunction   Immune-mediated myositis	Beydon et al.,32 Dalakas,30 Zhang et al.33	Favourable responses to steroids, IVIg and tocilizumab reported
Chronic neurological sequelae
Fatigue	Chronic neuroinflammation   Neuroendocrine dysfunction   Persistent respiratory and cardiac damage	Pandharipande et al.,34 Raman et al.,35 Mongioì et al.36	None
Cognitive impairment	Chronic neuroinflammation   Frontoparietal hypometabolism	Blazhenets et al.,37 Guedj et al.38	None, demonstrated to improve over months
Depression/altered mood	Stress (isolation, post-traumatic stress)	Rogers et al.39	No specific therapies proposed or tested for post-COVID-19 patients

PLEX = plasma exchange.

Acute neurological syndromes

Anosmia and ageusia were among the first focal neurological symptoms described in COVID-19, and generated interest in SARS-CoV-2’s potential neurotropism. Anosmia was reported in 5–35% of hospitalized patients, and may be higher among non-hospitalized patients with COVID-19. In some cases, it is the sole reported symptom, or persists far beyond the acute respiratory symptoms, negatively influencing the quality of life of survivors. Infection of the nasal mucosa and sustentacular cells with dissemination throughout olfactory nerve projections is one proposed mechanism of neuropathogenesis in COVID-19. Viral RNA has also been found in the olfactory bulb at post-mortem of some patients, although its association with neurological injury is not established.

Altered mental status is a commonly reported neurological finding associated with COVID-19 hospitalization. Abnormalities in EEG in COVID-19 related encephalopathy correlate with disease severity. Frontal slowing is the most common pattern observed and has been proposed as a biomarker for COVID-19 encephalopathy. Seizures are uncommon among COVID-19 patients; in a retrospective cohort of 1043 patients, 0.7% developed seizures in hospital and an even smaller proportion of those seizures occurred outside the context of pre-existing epilepsy although a recent retrospective study indicated that epileptiform abnormalities are frequently detected (48.7%) of hospitalized patients with COVID-19. Altered mental status, coma and seizures in COVID-19 are almost certainly multifactorial but can be stratified into metabolic/non-inflammatory versus inflammatory (e.g. encephalitis) categories.

Non-inflammatory encephalopathy

Many patients with COVID-19 and altered mental status (e.g. lethargy, confusion, coma) have clinical courses complicated by hypoxia, renal failure, electrolyte disturbances, sedating medications and underlying comorbidities. A third of critically ill patients with COVID-19 present with encephalopathy, often with frontal lobe-associated features,, either at the onset of illness or during the course of hospitalization, and encephalopathy is associated with increased mortality and poor functional outcome.,, Although encephalopathy has been reported in COVID-19 patients at all ages, patients beyond the sixth decade of life and those with pre-existing neurologic conditions (stroke, dementia, Parkinson’s disease) are most affected, particularly in the context of severe respiratory illness., While encephalopathy in the aforementioned circumstances is probably multifactorial, reports of patients with encephalopathy in the absence of severe respiratory illness suggest other possible mechanisms including bioenergetic failure and vascular dysfunction in SARS-CoV-2 infection. The association between encephalopathy and morbidity exists independently of respiratory disease, similar to that observed in sepsis-associated encephalopathy. Encephalopathy in non-COVID-19 patients is attributed to mitochondrial dysfunction, excitotoxicity and macro- or micro-ischaemic injury. One recent post-mortem analysis in New York revealed cerebral microthrombi in a subset of 67 hospitalized, deceased COVID-19 patients. Small, subcortical ischaemic events may result in confusion and cognitive dysfunction. Patients can also manifest multifocal cerebral microhaemorrhages or vascular leakage (Fig. 2) due to compromise of the cerebral endothelial cells., Radiographic findings in COVID-associated encephalopathy include non-specific white matter hyperintensities, diffusion restriction, microhaemorrhage and leptomeningeal enhancement., In one study of intensive care unit (ICU) patients with COVID-19 and encephalopathy, bilateral frontotemporal hypoperfusion was evident in all patients who underwent perfusion imaging for altered mental status, although this finding was disputed and has not been replicated. Surprisingly, brain MRI was normal in up to 46% of patients with COVID-19 and associated encephalopathy. In another study, patients with COVID-19 and cognitive impairment showed decreased metabolism in the fronto-parietal regions on fluorodeoxyglucose (FDG)-PET scans. Indeed, correlations between neuropathology and brain MRI findings (including normal imaging) have yet to be established.

Figure 2

Microvascular diseases with COVID-19. (A) Multiple congested blood vessels and microhaemorrhages are observed in the basal ganglia at post-mortem. (B) MRI of the same block of tissue shows hyper and hypointense signals corresponding to the blood vessels in A. The hyperintense signals represent fibrin clots while the hypointense signals are microhaemorrhages. (C) MRI of the pons shows similar punctate hypointense signals (arrows).

Inflammatory encephalitis

A minority of patients with COVID-19 encompass established diagnostic criteria for infectious encephalitis. There are convincing reports of encephalitis-like presentations associated with elevated levels of soluble IL-6, IL-18, TNF-α, CXCL10 and markers of glial and astrocyte activation in CSF.,,, Radiological findings associated with COVID-19 meningoencephalitis include mesial temporal lobe T2/FLAIR hyperintensities, varying from punctate to diffuse in subcortical white matter, the brainstem and claustrum, often accompanied by cerebral oedema. Case reports indicate that COVID-19 can present with an ADEM phenotype,, including oculomotor dysfunction, seizures and coma. Others have reported cases of acute necrotizing haemorrhagic encephalopathy that present initially with symmetric lesions in the thalami and are thought to be cytokine mediated. Some patients may present with isolated pseudotumor cerebri/benign intracranial hypertension presumably from meningitis., Opsoclonus-myoclonus syndrome, which has been observed in association with infections such as Epstein–Barr virus (EBV), chikungunya and Mycoplasma pneumoniae, has also been reported in patients with COVID-19, including those with mild respiratory disease. Most patients with COVID-19 and opsoclonus-myoclonus syndrome had partial recovery at 4 weeks after treatments including pulse steroids, intravenous immunoglobulin (IVIg) and anti-epileptic medications.

For presumed COVID-19 associated encephalitis, favourable therapeutic responses to corticosteroids and plasma exchange (PLEX) were observed in a subset of patients, although factors predicting a beneficial therapeutic response remain to be defined., Whether the neuroinflammation observed clinically, neuroradiologically and neuropathologically is due to direct viral invasion, para-infectious or autoimmune processes remains unknown. In most COVID-19 cases with encephalitis, SARS-CoV-2 RNA is not detectable in CSF via PCR with reverse transcription (RT-PCR), favouring an immune-mediated mechanism of disease.,,,

Cerebrovascular disease

Patients with COVID-19 have an increased rate of stroke compared to other disease cohorts, with higher NIHSS scores compared to non-COVID-19 associated stroke. Over half of strokes among patients with COVID-19 are cryptogenic, with a higher proportion of large vessel occlusions. Some series have reported higher than expected rates of posterior circulation strokes (35.3%)., Hypercoagulability induced by systemic and focal inflammation has been implicated in COVID-19 associated strokes that include both arterial and venous thromboembolic events. Cerebral venous sinus thrombosis (CVST) among patients with COVID-19 can also occur with an abnormally activated prothromboplastin time (aPTT) and elevated D-dimer levels., COVID-19 associated CVST has an estimated in-hospital mortality of 40% in a cohort that included non-ventilated patients. In fact, CVST represents 4% of cerebrovascular complications in COVID-19 with an estimated frequency of 0.08% among hospitalized patients. In comparison, CVST accounts for 0.5–1% of all strokes among non-COVID-19 patients and occurs in ∼2–5 per million people each year (0.0002–0.0005%). Elevated D-dimer levels are more common among COVID-19 patients presenting with both ischaemic and haemorrhagic stroke and are associated with higher all-cause mortality. As there is an increased risk of thromboembolism during COVID-19, multiple studies have compared standard dose thromboprophylaxis to high and intermediate-dose prophylactic anticoagulation in hospitalized patients with COVID-19. The largest of these trials, INSPIRATION randomized clinical trial found no difference in all-cause mortality or venous thromboembolic events in patients treated with standard versus intermediate-dose thromboprophylaxis in critically ill patients with COVID-19, in contrast to earlier retrospective studies showing potential benefit in ICU patients. Interim unpublished results of the multiplatform merged randomized control trial (mpRCT) ATTACC/REMAP-CAP/ACTIV-4A found similar results in the critically ill cohort as well as a signal towards harm with therapeutic anticoagulation. Interestingly, this study found improved survival in moderately ill patients with COVID-19 treated with intermediate-dose anticoagulation, suggesting severity of disease may be important in determining appropriate thromboprophylaxis in hospitalized patients., Further studies are required to determine whether prophylactic anticoagulation specifically reduces the risk of stroke in COVID-19, particularly because of reports of haemorrhagic stroke in hospitalized patients while receiving therapeutic anticoagulation. Indeed, the risk of haemorrhagic stroke is higher than predicted among COVID-19 patients and is associated with elevated serum ferritin. Similarly, microhemorrhages and acute haemorrhagic necrotizing encephalitis have been reported in patients with COVID-19. Outcomes including risk of death and duration of hospitalization following intracerebral or subarachnoid haemorrhages are worse among patients with COVID-19.

Recent reports of vaccine-induced thrombotic thrombocytopaenia (VITT) following administration of adenovirus vector-based COVID-19 vaccines have raised concern. As of 4 April 2021, there had been 169 cases of VITT-CVST reported to the European Medicines Agency out of 34 million doses administered of the ChAdOx1 nCoV-19 (AstraZeneca) vaccine, with an incidence of VITT estimated at 1 per 100 000 exposures. The reported rate of VITT-CVST after administration of 6.86 million doses of the Ad26.COV2.S adenoviral vector vaccine (Johnson & Johnson/Janssen) was 0.87 cases per million (0.000087%)., Most cases occurred in females <50 years of age, and most patients displayed high levels of antibodies to platelet factor 4 (PF4)-polyanion complexes, prompting comparisons to heparin-induced thrombotic thrombocytopaenia. The precise pathophysiology of VITT is unknown to date; simultaneous thrombosis and thrombocytopaenia in VITT has only been reported for adenoviral vector vaccines although there have been five possible reports of CVST with normal platelet counts among 4 million doses of the Moderna mRNA vaccine. Among 54 million doses of the Pfizer-BioNTech mRNA vaccine, there have been 35 reports of CNS thrombosis without thrombocytopaenia (0.00006%). CSVT is a serious but rare condition associated with SARS-CoV-2 vaccination, but there remains a consensus among health authorities that the benefits of widespread vaccination outweigh the potential risks, particularly when one considers the rate of thrombosis in COVID-19 infection.

Acute PNS disorders

Autoimmune polyradiculoneuropathies such as GBS or Miller-Fisher syndrome have been reported in patients with SARS-CoV-2 infection, with and without respiratory symptoms. These disorders can be triggered by systemic infections and have been reported in patients with other coronavirus infections such as Middle East respiratory syndrome (MERS) and SARS-CoV-1. Most case reports of GBS in COVID-19 describe the common syndrome of ascending weakness, areflexia with supporting CSF and nerve conduction studies, and are of the acute inflammatory demyelinating polyneuropathy (AIDP) type. Disease onset is between 5 and 10 days after acute COVID-19 symptoms (including anosmia, respiratory and gastrointestinal symptoms), which in the ICU settings helps distinguish GBS from critical illness neuropathy that appears later in disease course.,, Patients with COVID-19-associated GBS respond to standard treatments (e.g. IVIg, PLEX) although how COVID-19 affects treatment responsiveness remains uncertain. Of note, a UK epidemiological cohort study showed rates of GBS have fallen during the current pandemic, probably resulting from increased public health efforts that have reduced transmission of more common infectious triggers.

Myalgia and weakness occur in 30–50% of hospitalized patients with COVID-19,, and are frequently reported by non-hospitalized patients. While myalgia is a common symptom during many viral illnesses, the mechanism by which SARS-CoV-2 infection causes debilitating muscle pain and weakness is unknown. Myositis and rhabdomyolysis as a complication of COVID-19 are well recognized with elevated serum creatine kinase (>10 000) levels as a common finding which correlates with mortality in hospitalized patients. There have also been multiple reports of muscle oedema demonstrated on MRI., While other viruses such as influenza are known to directly invade skeletal myocytes in vitro, to date there is no evidence for infection of skeletal myocytes with SARS-CoV-2., Myositis in COVID-19 could be triggered by host immune responses to the virus. Muscle biopsies from COVID-19 patients show perivascular inflammation including a case of type 1 interferonopathy associated myopathy in a young patient with SARS-CoV-2 infection. There are reports of COVID-19 patients with elevated creatine kinase levels and muscle weakness who respond to immunosuppression including high dose glucocorticoids as well as IVIg, prompting comparisons to immune-mediated myositis. There is also a recent report of a patient with proximal and bulbar weakness in COVID-19 with positive anti-SSA and SAE-1 antibodies who was successfully treated with the humanized monoclonal antibody against the IL-6 receptor, tocilizumab. While these neurological syndromes are observed in the acute setting, they have the capacity to exert long-term effects, as described next.

Patients who develop severe COVID-19 pneumonia often require prolonged ICU care. As expected, critical illness polyneuropathy (CIP) and myopathy (CIM) have been reported as complications of SARS-CoV-2 infection. While the pathophysiological mechanisms underlying CIM and CIP are unknown, both disorders are assumed to result from microcirculatory and metabolic changes brought on by severe physiological stress. Based on electrophysiological and pathological studies, there is no evidence that COVID-19 associated CIP/CIM has distinctive features, and treatment to date has been supportive. In fact, the lasting neurological consequences of prolonged hospitalization with or without intensive care and the associated interventions for patients with COVID-19 remain unclear.

Chronic neurological sequelae

The long-term neurological impact of COVID-19 is uncertain, but it is already apparent that a range of signs and symptoms emerge among patients hospitalized with COVID-19 while non-hospitalized patients also exhibit neurological disorders that arise after the acute COVID-19 illness phase (Fig. 3). The lingering or delayed neurological syndromes have been termed long COVID or post-acute sequelae of SARS-CoV-2 (PASC) and are composed of a wide range of symptoms and signs including neurocognitive symptoms with associated impaired performance on neuropsychological testing. Of note, neurocognitive and mood alterations among ICU survivors are well recognized phenomena, often attributed to sedating medications as well as systemic inflammation and neuronal injury. Notably, these ICU-related effects can confound the evaluation of chronic sequelae among survivors of severe acute COVID-19. A study evaluating patients with COVID-19 at 2–3 months post-hospitalization (approximately a third of patients required ICU) reported that those patients reported significantly higher rates of depressive symptoms and decreased quality of life compared to age- and comorbidity-matched controls. Moreover, abnormalities in visuospatial and executive function were detected among COVID-19 survivors compared to controls when assessed by the Montreal Cognitive Assessment tool (MoCA), recapitulating clinical experience of patients with post-COVID-19 who report apathy, short-term/working memory difficulties and ‘brain fog’ after SARS-CoV-2 infection., A recent study of patients post-COVID-19 without hospitalization reported ‘brain fog’, headache, anosmia, dysgeusia and myalgia as the predominant persisting symptoms. Over half of hospitalized COVID-19 patients report significant fatigue months after discharge, particularly among those who required admission to the ICU. Similarly, persistent psychological distress is reported by half of hospitalized patients with COVID-19-related ICU admission as well as those COVID-19 patients not requiring the ICU. A retrospective cohort analysis of over 200 000 patients in the UK found that 12.8% of patients with COVID-19 received a new neurological or psychiatric diagnosis in the 6 months after initial infection. In the same study, nearly half of ICU-COVID-19 survivors had a neurological or psychiatric illness at 6-month follow-up, of which half were new diagnoses. Of note, frontotemporal FDG hypometabolism reported for acute COVID-19, discussed previously, was also observed among COVID-19 patients with cognitive symptoms >3 weeks after initial illness, accompanied by brainstem and thalamus hypometabolism in ‘long COVID’ patients, compared to controls. A separate study of eight patients in the subacute and chronic stages of recovery from COVID-19 observed a similar pattern of bilateral frontoparietal hypometabolism, which resolved at the 6-month follow-up assessment and was accompanied by improved MoCA scores. FDG-PET imaging is a potentially useful research tool although it is not validated for diagnosis of COVID-19 related neurocognitive impairments, which require clinical evaluation. Future studies of cognitive impairment in COVID-19 survivors must take into account the fact that hospitalization for any infection is associated with an increased 10-year risk of dementia, particularly vascular dementia and Alzheimer’s disease.

Figure 3

Chronic neurological sequelae of COVID-19. Several long-term neurological syndromes result from SARS-CoV-2 among hospital- and community-treated patients, termed long COVID or post-acute sequelae of COVID-19 (PASC). These syndromes include neurocognitive, mood and sleep disorders, dysautonomia, diverse pain syndromes, as well as marked exercise intolerance and fatigue. These protracted syndromes remain to be fully defined in longitudinal cohort studies.

Patients with COVID-19 also develop autonomic instability that manifests as tachycardia, postural hypotension, hypertension, postural orthostatic tachycardia syndrome, low-grade fever with associated bowel, bladder or sexual dysfunctions., Cardiac MRI of COVID-19 survivors at 2–3 months after symptom onset showed evidence of fibrosis and inflammation, which was correlated with serum inflammatory markers (e.g. CRP, calcitonin), possibly accounting for the exercise intolerance reported by patients.

The spectrum of symptoms described in long COVID has prompted comparisons with myalgic encephalomyelitis or chronic fatigue syndrome (ME/CFS). Indeed, the overlap in symptoms between post-acute COVID-19 syndromes and ME/CFS is remarkable for the shared symptomatology including fatigue, autonomic instability, post-exertional myalgia or weakness as well as neurocognitive impairments.,, Nonetheless, other viral illnesses (e.g. Dengue, West Nile disease, mononucleosis) are also associated with substantial disabilities that resemble the previous symptom complex. The precise diagnosis and management of neurological symptoms in long COVID is an emerging area of study, which is in evolution as more studies become available. Important caveats in considering persistent or delayed neurological disorders related to COVID-19 include the contribution of comorbid illnesses and their associated therapies to neurological disease as well as the potential for uncovering previously unrecognized illnesses.

Laboratory analyses of nervous system tissues and fluids

Analyses of CSF from patients with COVID-19 vary widely depending on the associated neurological disorder although pleocytosis, especially lymphocytic, and elevated protein are common findings, particularly among patients with other features of encephalitis. The IgG index is increased in many patients with COVID-19 together with the presence of antiviral and antiviral receptor (e.g. ACE2) antibodies, indicative of intrathecal synthesis., In contrast, viral RNA is infrequently detected in CSF using standard RT-PCR protocols,, although the timing of the CSF collection in relation onset is often not reported. Host innate immune responses were also apparent in CSF from patients with COVID-19 based on reports of neopterin and β2-microglobulin detection in CSF. Similarly, several chemokines and cytokines in CSF have shown to be associated with COVID-19-related neurological disease (e.g. encephalitis) including IL-8, TNF-α, IL-6 as well as neural cell type-specific markers (e.g. GFAP, neurofilament and tau). However, a specific diagnostic profile in CSF for COVID-19 associated neurological disease awaits definition. Antibodies associated with autoimmune encephalitis have been reported concurrently with SARS-CoV-2 infection, including anti-GD1b, -NMDA-R,, and -CASPR2. While these reports are intriguing, a direct link between SARS-CoV-2 infection and the development of these autoantibodies has not been established. Interestingly, there are emerging reports of non-neurological autoimmune disorders including psoriatic arthritis, rheumatoid arthritis and immune thrombocytopenic purpura developing after COVID-19. Possible explanations for this phenomenon include transient immunosuppression during acute viral illness, including suppression of regulatory T and B cells resulting in impaired self-tolerance, as has been suggested in other viral infections. In susceptible individuals, the process of immune reconstitution following COVID-19 may ‘unmask’ autoimmune conditions, including multiple sclerosis and neuromyelitis optica spectrum disorders., In contrast, other groups have proposed that T-cell exhaustion might contribute to autoimmune neuropathogenesis in COVID-19.

As with CSF studies, autopsy-based neuropathological findings are diverse. Several variables need to be considered in interpreting the neuropathological findings including the presence and severity of prior or concurrent comorbidities, duration in ICU and ventilator support, concomitant therapies and the circumstances of death. Moreover, for many neuropathological reports of COVID-19, a corresponding clinical phenotype was not observed or reported. Nevertheless, reports range from the findings of absent neuropathology to hypoxic/ischaemia changes, acute infarction and haemorrhagic lesions with endotheliitis. ADEM- and ATM-like findings have been observed in select cases., Post-mortem studies of patients with ADEM-associated COVID-19 report periventricular inflammation, characterized by foamy macrophages and axonal injury., Conversely, other neuropathological studies have identified lymphocyte-predominant inflammation in the meninges, brainstem and perivascular spaces with significant neuronal and axonal loss. Meningoencephalitis, haemorrhagic posterior reversible encephalopathy syndrome, as well as diffuse leukoencephalopathy and microhaemorrhages have also been reported.,, While a number of post-mortem studies indicate there is a paucity of immune cell infiltration within the neuroaxis,,, recent studies have found marked microglial activation and CD8+ T cells in the brainstem and cerebellum., In fact, one study reported pan-encephalitis in a cohort of patients with severe pulmonary-associated COVID-19. Microscopy in larger studies (n = 43) have described diverse findings including astrogliosis with activated microglia and infiltrating T cells in brain parenchyma, together with ischaemic lesions in a subset of patients.,, In one post-mortem study using imaging mass cytometry, distinct neuropathological features within the brainstem and olfactory bulb of COVID-19 patients were identified, including microglial nodules, CD8+ T-cell infiltration, and increased ACE2 expression in blood vessels. These findings were not as pronounced in control patients who had been on ECMO but did not have COVID-19. Nevertheless, some authors have commented that collectively the neuropathological findings, especially microglia activation in COVID-19 resemble that observed in patients with hypoxia and sepsis.,

Mechanisms of neurological disease

Multiple putative mechanisms of disease have been proposed for COVID-19 induced nervous system disorders including coagulopathies as well as virus-associated host responses. Indeed, it is probable that specific pathogenic processes underlie the individual neurological presentations associated with COVID-19 in both the CNS (Fig. 1A) and the PNS (Fig. 1B). We review the different proposed mechanisms next.

Cerebrovascular disease/bioenergy failure

Microvascular injury characterized by thinning of the basal lamina of endothelial cells, fibrinogen leakage and microhaemorrhages has been described in the brainstem and olfactory bulb of deceased COVID-19 patients corresponding to visible MRI changes. These observations are also complemented by other neuroimaging studies in which cerebral infarction was the most common finding on conventional brain MRI. Most post-mortem analyses have shown signs of thrombotic microangiopathy and endothelial injury with minimal evidence of prototypic vasculitis. This pattern is suggestive of endotheliitis. Although there have been several case reports of CNS vasculitis associated with COVID-19, none have confirmed the diagnosis histologically. A cohort of patients with stroke and COVID-19 in Wuhan, China, showed elevated serum levels of IL-6, IL-8 and TNF-α, a finding that has been replicated in several subsequent studies. Both IL-8 and TNF-α promote the release of von Willebrand factor, a marker of endothelial damage that is elevated in both ICU and non-ICU patients with COVID-19, while IL-6 inhibits cleavage of von Willebrand factor leading to accumulation of multimers that promote platelet aggregation. These changes are bolstered by findings of damaged cerebral blood vessels or endotheliitis that was associated with extravasation of fibrinogen. These mechanisms of disease are highly plausible because of the frequency of coagulation-related events during COVID-19. Indeed, neuroimaging studies point to abnormal energy metabolism, shown by reduced FDG detection in frontal lobes of patients with acute COVID-19.

Viral neuroinvasion

SARS-CoV-2 infects respiratory cells via engagement of the angiotensin-converting enzyme 2 (ACE2) receptor,,, with a higher binding affinity than other coronaviruses such as SARS-CoV-1. The ACE2 receptor is present on type II alveolar and respiratory epithelial cells, cardiomyocytes, neurons and astrocytes., This receptor is also present in pericytes and smooth muscle cells of cerebral blood vessels and is expressed in the thalamus, cerebellum and brainstem nuclei of humans. After binding to ACE2, cleavage of the spike (S) protein of SARS-CoV-2 by transmembrane serine protease 2 (TMPRSS2) facilitates cell entry. Alternative docking receptors including neuropilin-1 (NRP1) and basigin (BSG)/CD147 are found at higher levels in the CNS. Similarly, alternative proteases including furin and cathepsin might permit viral entry in cells with low levels of TMPRSS2 expression (e.g. brain).

Several anatomic routes of neuroinvasion by SARS-CoV-2 have been proposed. The integrity of the blood–brain barrier is compromised in multiple conditions associated with mortality in COVID-19, including hypertension, diabetes, smoking and stroke. Areas of increased vascular permeability or lack of blood–brain barrier, such as the pituitary and median eminence of the hypothalamus are also rich in ACE2, NRP1 and TMPRSS2, thus representing possible portals of entry into the CNS. SARS-CoV-2 infects nasal epithelium and perhaps olfactory bulb cells, presenting another entry portal to the CNS, as suggested for other coronaviruses., A recent post-mortem analysis of humans with COVID-19 detected SARS-CoV-2 by RT-PCR in neuroepithelium, the olfactory bulb, trigeminal ganglion and brainstem, albeit at low levels. Interestingly, olfactory nerves terminate in the frontal cortex as well as the hypothalamus and amygdala, structures that are implicated clinically, radiographically and electrographically in the neurological sequelae of COVID-19. The importance of the choroid plexus in the development of COVID-19 associated neurological disease in conjunction with neuroinflammation has been highlighted recently in a large study predicated on RNA deep sequencing of brain-derived single cell nuclei transcriptomes. The lack of evidence for productive infection of trafficking immune cells by SARS-CoV-2 to date makes a Trojan horse mechanism of neuroinvasion less likely. Nonetheless, viral proteins and RNA have been detected in CD68+ macrophages isolated from bronchoalveolar lavage of COVID-19 patients. SARS-CoV-2 RNA levels in brain tissue detected by RT-PCR are low and seemingly independent of the presence or absence of apparent neurological dysfunction and histopathological alterations., Immunodetection of SARS-Cov-2 viral antigens in neurons from autopsied patients with COVID-19 underscores the potential for direct viral invasion as an important disease determinant.

Remdesivir, a nucleoside analogue that inhibits RNA-dependent replication of SARS-CoV-2, is the only direct antiviral agent approved for COVID-19 treatment despite preliminary results showing no impact on mortality or progression to mechanical ventilation. Molnupiravir is orally available nucleoside analogue that induces coronavirus lethal mutagenesis and is in phase 2 and 3 trials for treatment of COVID-19. A recent randomized control trial of the TMPRSS2 inhibitor, camostat mesylate, in hospitalized patients with COVID-19 did not have any impact on recovery, progression to ICU or mortality.

Host neuroimmune responses

Post-infectious neuro-inflammation triggered by expression of viral antigens into the CNS is another proposed mechanism of encephalitis in COVID-19. While human data supporting this hypothesis are limited, a recently published study using a murine model showed a subunit of the SARS-CoV-2 spike protein (S1) crosses the BBB via absorptive transcytosis when administered intravenously and intra-nasally. Indeed, neuropathological studies demonstrate glia activation and occasional leucocyte infiltrates in patients with COVID-19 although the associated molecular pathways (e.g. cytokine, protease, or free radical release) induced are unclear. CSF studies suggest activation of innate immune responses with elevated levels of β2-microglobulin and neopterin and the presence of dedifferentiated monocytes., This is associated with increased levels of neurofilament suggesting neuronal injury. Autoimmune mechanisms including both antibody- as well as cell-mediated immune injury of neural tissue are also plausible, given the recognition of autoimmune processes in the systemic COVID-19 pathogenesis. The injury and loss of endothelial cells in arterioles, venules and capillaries represents another neuropathogenic avenue via disruption on the blood–brain barrier and through endothelia production of immune molecules in the lung, kidney and heart of patients with COVID-19. These latter events can be initiated by systemic immune activation as well as a coagulation diathesis. An important qualification to these mechanisms is that concurrent clinical events including systemic hypoxia-ischaemia might affect immune processes within the nervous system. Among patients with COVID-19 associated cerebrovascular disease, autoimmune processes have been directly implicated. For example, the contribution of antiphospholipid antibodies to ischaemic stroke in patients with COVID-19 is controversial. Zhang et al. described three COVID-19 patients with coagulopathy and multi-territory infarcts and anticardiolipin and anti-β2 microglobulin antibodies. Subsequent studies have reported lupus anticoagulant positivity in more than half of COVID-19 patients. Most case reports of antiphospholipid antibodies in COVID-19 do not include repeat assays 12 weeks apart, which is required for the diagnosis of antiphospholipid antibody syndrome. Transient elevation of lupus anticoagulant during systemic inflammation is common, and several infections are associated with false positive antiphospholipid assays, including HIV, hepatitis C virus and syphilis, making current reports of antiphospholipid antibodies in COVID-19 difficult to interpret.

Similarly, autoimmunity is also incriminated in COVID-19 associated GBS; anti-ganglioside antibodies implicated in autoimmune polyradiculoneuropathies such as anti-Gq1b, -GM1 and -GD1b antibodies have been reported in patients with COVID-19 presenting with cranial neuropathies, weakness, areflexia and sensory ataxia. Anti-ganglioside antibodies are most strongly associated with more aggressive axonal motor neuropathies and poorer functional outcomes compared to AIDP. The rare presence of these antibodies raises concern about potential molecular mimicry mediated by SARS-CoV-2 that could trigger autoimmune responses with important implications for vaccine safety. The spike (S) protein of SARS-CoV-2 is highly glycosylated; thus, the development of anti-glycan antibodies may be essential for an effective host immune response in COVID-19. In a microarray study of 800 human carbohydrate antigens, levels of anti-glycolipid antibodies associated with GBS, including GM1a, GD1a and GD1b significantly higher in COVID-19 patients compared to healthy controls. In this latter study, there was no direct correlation with antibody titre and clinical features of GBS. Anti-glycan antibodies are also observed in other viral and bacterial infections (HIV, EBV, Neisseria meningitidis,) as well as autoimmune diseases such as Crohn’s disease, and thus may merely be a marker of systemic inflammation. Of relevance, there were no reported cases of GBS in the three major COVID-19 vaccine trials.

While randomized control trials demonstrate dexamethasone and tocilizumab improve respiratory outcomes in hospitalized patients, their effects on neurological disease in COVID-19 is presently supported only by case reports., A subset of COVID-19 associated encephalopathies are responsive to steroids and IVIg, and there is a single report of a young patient with encephalitis and SARS-CoV-2 (based on CSF lymphocytosis and T2/FLAIR hyperintensities on MRI), which resolved after treatment with IVIg and tocilizumab. In most cases with a positive response to immunosuppressive or modulatory therapy, SARS-CoV-2 was not detected in CSF, further supporting a para-infectious/immune-mediated basis for disease.

Future perspectives

Given the mounting impact of SARS-CoV-2 infection globally together with the increasing recognition of associated neurological disorders, it is imperative to define the types of COVID-19 related neurological syndrome, including those caused directly by viral infection versus those arising from systemic illness, the impact of different viral variants on neurological disease, as well as identifying informative diagnostic tools and effective therapies. GWAS studies have identified susceptibility genes for severe respiratory illness with COVID., Similar studies to identify host factors associated with neurological complications would also be useful. The long-term neurological sequelae of COVID-19 remain unclear and await delineation in longitudinal studies. The neurodevelopmental impacts of COVID-19 are also unknown in utero as well as in infants or adolescents; this issue could have substantial lasting effects that require further investigation. Finally, a more comprehensive understanding of the pathogenic mechanisms underpinning the neurological syndromes associated with COVID-19 will advance therapeutic options for affected patients.

Literature search strategy and selection criteria

Studies were selected from the peer-reviewed literature using NCBI and Google Scholar. We searched the databases using the following keywords: central and peripheral nervous systems, COVID-19, SARS-CoV-2, coronavirus, stroke, encephalopathy, neurocognitive impairment, hypercoagulability, encephalitis, neurologic infection, seizure and neuroinflammation. We also reviewed bibliographies of relevant articles. Non-peer-reviewed studies and single case reports were not included as references unless they were highly informative.