Potential routes and mechanisms of CNS involvement in COVID-19

The mechanisms of CNS infection in human coronaviruses (HCoV-OC43, HCoV-229E, MERS-CoV) have not yet been definitively elucidated. However previous studies demonstrated the presence of SARS-CoV in brain samples by immunohistochemistry, electron microscopy and real-time reverse transcription-polymerase chain reaction (RT-PCR) [12, 13] and in cerebrospinal fluid (CSF) by polymerase chain reaction (PCR) in the acute phase of patient illness [12]. Studies using experimental animals established that SARS-CoV may enter the brain via the olfactory nerve and initially spread to connected brain regions before proliferating more widely – including to areas such as the brainstem and cardiorespiratory centre in the medulla; the latter may potentially contribute to death. The infection of the olfactory nerve is followed by involvement of regions such as the piriform and infralimbic cortices, basal ganglia and dorsal raphae nuclei in the midbrain [14]. Other regions, including the thalamus and hypothalamus are less consistently positive whereas there is significant neuronal loss in the cingulate and anterior olfactory nuclei; a pattern of distribution suggestive of trans-neuronal spread. The common presentation of altered or loss of smell (hyposmia) in COVID-19 suggests this transmission via the olfactory nerve and its anatomical brain connection pathways may also to be utilized by SARS-CoV-2 [14].

Another interesting commonly reported clinical symptom of COVID-19 is hypogeusia (loss of taste) [7]. The sense of taste is conducted via the facial nerve (from the anterior two-thirds of the tongue) and glossopharyngeal nerve and superior laryngeal branch of the vagus nerve (from the posterior two-thirds of the tongue), the information is passed to the nucleus of the solitary tract in the brainstem and then to the thalamus. Thus, these cranial nerves potentially provide another route for trans-neuronal spread to the brain [2].

The presence of SARS-CoV-2 in the general circulation could provide a haematogenous route of entry to the CNS. The virus might utilize the ACE2 receptors in the endothelial cells followed by subsequent budding of the viral particles from the capillary endothelium, damaging them, gaining access through the blood–brain barrier (BBB) and initiating viral budding through interaction with ACE2 receptors in the neurones. Varga et al. claim to demonstrate SARS-CoV-2 particles by electron microscopy within endothelial cells in the blood vessels of postmortem tissue from kidney, lung, heart and liver (associated with lymphocytic endotheliitis) [15]. The infection of endothelial cells causes rupture of the capillaries leading to bleeding or haemorrhagic infarction, which have been recently reported in COVID-19 patients (discussed further below) and can result in a fatal outcome.

The third possible route of entry of SARS-CoV-2 into the CNS is via immune cells. The virus enters the body through the respiratory tract and first infects the epithelial cells of the trachea and bronchi and the alveolar cells of the lung, subsequently infecting the resident immune cells which carry the virus to other organs, including the brain [16].

Clinical evidence of CNS involvement by SARS-CoV-2

SARS-CoV-2 infection presents with common symptoms of fever, cough, fatigue, headache and dyspnoea. In severe cases, patients develop pneumonia, acute respiratory distress syndrome, cardiac symptoms and multi organ failure. Recent evidence suggests that COVID-19 may also involve the central and peripheral nervous systems (PNS). There are increasing clinical reports of symptoms suggesting CNS involvement; such as headache, nausea, vomiting, impaired consciousness and acute cerebrovascular disease including stroke, venous sinus thrombosis and intracerebral haemorrhage [5-7, 23].

In a retrospective study of 214 patients with COVID-19 by Mao et al., 36.4% of patients were found to have non-specific neurological manifestations, such as dizziness, headache, nausea and more specific symptoms such as hypogeusia and hyposmia [7]. In severe cases with acute cerebrovascular disease or disturbance of consciousness many patients showed lower lymphocyte levels compared with those without CNS symptoms. The CNS involvement appeared to indicate poor prognosis and death [7].

Wang et al. reported probable neurological manifestations such as headache (6.5%), vomiting (3.6%), dizziness (9.4%), nausea (10.1%) and myalgia (34.8%) in their review of clinical manifestations in 138 hospitalized COVID-19 positive patients [5]. Additionally

Chen et al. studied 99 SARS-CoV-2 positive patients with common symptoms of fever, cough and shortness of breath in the majority of patients, and described neurological manifestations in a number of them – specifically, nausea and vomiting (1%), confusion (9%), headache (8%) and muscle ache (11%) [23].

More specific neurological presentations of meningo-encephalitis have been described by several authors in different age groups, including in children [24]. A 24-year-old male with meningitis/encephalitis associated with COVID-19 was described by Moriguchi et al. [25]. The patient was admitted with unconsciousness and convulsions after complaining of fever and vomiting but had no neck stiffness. Brain MRI showed hyperintensity along the lateral ventricle and hippocampus. Additionally, Huang et al. and Xiang et al. published cases in which SARS-CoV-2 RNA was isolated from the CSF, suggesting neuro-virulence of the virus [26, 27], however, it is not certain if this indicates replicative/infective potential of the virus.

Another neurological entity reported in COVID-19 patients has been termed acute necrotizing encephalopathy. It is reported by Dixon et al. in a 59-year-old female who deteriorated on day 6 after admission. MRI demonstrated brain stem swelling and symmetrical haemorrhagic lesions in the brain stem, amygdalae, putamina and thalamic nuclei [28]. Cases with a similar presentation and characteristic MRI features including symmetric, multifocal lesions with invariable thalamic, brain stem, cerebral white matter and cerebellar multiple necrotic regions were also reported by Mao et al. and Poyiadji et al. [7, 29].

A review of the current literature showed that acute cerebrovascular disease is one of the more common, serious neurological complications seen in COVID-19 populations. In Italian, Dutch and Chinese cohorts of COVID-19 patients admitted with confirmed infection, the rate of ischaemic stroke was 2.5, 3.7 and 5% respectively, despite venous thromboprophylaxis [8, 30, 31].

Mao and colleagues reported cerebrovascular accidents (CVA) in 5% of their cohort; with five ischaemic and one haemorrhagic stroke [7]; two of these patients were admitted with acute cerebrovascular accident as the initial presentation without prodromal symptoms but were found to have lung lesions and tested positive for COVID-19. The other patients presented with cough, fever and acute cerebrovascular symptoms, the COVID-19 test was negative initially but confirmed positive on repeat a few days later.

CVA resulting in an intracranial bleed is reported by Sharifi-Razavi et al. in a 79-year-old SARS-CoV-2 positive male who presented with fever, cough and coarse crepitation in the base of the lung, where the lung CT showed a ground-glass opacity in the left lower lobe [32]. There were no co-morbidities or any known predisposing factors for intra cerebral haemorrhage, such as hypertension or anticoagulant treatment, and results were normal for platelets and prothrombin time.

On the other hand, Helms et al. reported acute ischaemic strokes (diagnosed by MRI) in SARS-CoV-2 positive patients with encephalopathy who presented with prominent agitation and confusion and corticospinal tract signs; they suggested that higher D-dimers may be indicative of a higher risk of cerebrovascular disease (D-dimer is a product of fibrin clot degradation and indicates recent or current clot formation) [33].

Interestingly, CVA in COVID-19 is not only confined to older patients or those with cardiovascular co-morbidities, but it also has led to younger patients presenting with ischaemic stroke, including large vessel occlusions [34]. Additionally, COVID-19 patients can develop significant hypoxia leading to decreased cerebral oxygenation and infarcts, particularly in those with pre-existing cerebrovascular disease [35, 36].

Many authors have described the involvement of the spinal cord and peripheral nervous system in COVID-19 patients presenting with Guillain–Barre Syndrome, acute myelitis and encephalomyelitis [29, 37-39]. A possible association between CNS focal symptoms compatible with demyelinating disease (in the spinal cord) and SARS-CoV-2 infection was recently reported by Domingues et al. [40]. Non-specific muscle-related symptoms were reported by Mao et al. and patients showed elevated creatine kinase and lactate dehydrogenase levels compared to those without muscle symptoms [7]. Although it is possible that this could be related to ACE2 in skeletal muscle [41], other causes may be critical illness myopathy; an infection-mediated harmful immune response due to elevated pro-inflammatory cytokines.

Finally, some non-specific psychological problems are also recorded in COVID-19 patients, such as anxiety, depression, insomnia and distress [6, 42] and some patients with COVID-19 were found to be mentally confused [23]. Xu et al. found that 35% had mild, and 13% had moderate to severe psychological symptoms when they reviewed 89 survivors of COVID-19 [42].

Postmortem studies of COVID-19

Recently, a number of articles have reported postmortem examination of COVID-19 patients [43-47]. The pathological features of COVID-19 greatly resemble those seen in SARS and MERS coronavirus infection [12, 13, 48]. The most common postmortem findings are diffuse alveolar damage, acute renal tubular injury, lymphocyte depletion and evidence of micro- or macrothrombosis in many organs. Vasquez-Bonilla et al. conducted a systematic meta-analysis of histopathological observations from patients who died with COVID-19 [49]. They concluded that SARS-CoV-2 infection can result in diverse, multi-organ pathology, the most significant being; diffuse alveolar damage, microthrombi, bronchopneumonia, necrotizing bronchiolitis, lymphocytic myocarditis, acute tubular injury and haemophagocytosis and histiocytosis in the lymph nodes and bone marrow. The reported pathology in the brain includes microthrombi, ischaemic necrosis, acute haemorrhagic infarction and oedema.

In a recent publication on autopsy findings in the United Kingdom by Hanley et al., diffuse alveolar damage was the most consistent lung finding, with thrombotic phenomena in at least one major organ, lymphocyte depletion (particularly T-cells) in haematological organs and haemophagocytosis [50]. There were additional, unexpected findings such as acute pancreatitis, adrenal microinfarction, pericarditis, disseminated mucormycosis, aortic dissection and marantic endocarditis. These findings support the evidence of there being four dominant inter-related pathological processes in severe COVID-19; diffuse alveolar damage, thrombosis, haemophagocytosis and immune cell depletion [50]. Brain examination was also carried out in this cohort by the authors of this review (see below for more details). However, in general, documentation of detailed brain pathology in COVID-19 patients is incomplete.

In their articles detailing postmortem findings in patients who died with COVID-19, Xu et al. and Fox et al. did not examine the brain [46, 47]. Barton et al. reported no macroscopic abnormalities in the brain but microscopic examination was not undertaken [44] and Paniz-Mondolfi et al. reported only electron microscopy of brain tissue, claiming demonstration of viral particles in small vesicles within endothelial cells [51], however, this requires confirmation by further studies. In contrast, Schaller et al. report no morphologically detectable pathology and no signs of encephalitis or vasculitis in their cohort of 10 patients [52].

Brains from 18 patients who died after the onset of symptoms of COVID-19 were examined by Solomon et al. [36]. They showed ischaemic changes and reactive changes of Alzheimer type II astrocytes but there was no specific pathology such as meningitis or encephalitis. Immunohistochemical staining for SARS-CoV-2 in the above study was negative but the virus was detected at low levels by RT-PCR in five patients, however, there was no correlation with these results and the interval from the onset of symptoms to death.

More specific brain pathology was reported by Reichard et al. in a 71-year-old male previously diagnosed with ischaemic heart disease who developed postoperative complications including pneumonia and who was subsequently proven to be positive for SARS-CoV-2 [53]. The patient died after more than 2 weeks of hospitalization and the postmortem gross examination of the brain showed widespread recent ischaemia and small foci of intraparenchymal bleeding associated with damage to the white matter resulting in axonal injury, demyelination and macrophage infiltration. The authors comment on the resemblance of the pathology to that of acute disseminated encephalomyelitis (ADEM) although it is not typical and also raises the differential diagnosis of the cause being vascular in origin; possibly caused by microthromboembolic events related to COVID-19. These appearances are similar to those reported by Jaunmuktane et al., Bryce et al., and Remmelink et al. who also described cases with large and small infarcts, multiple small subcortical infarcts, foci of ischaemic necrosis and small microbleeds [54-56].

The authors of this review examined the brain from eight cases obtained as the first sequentially available consented autopsies from individuals who had died of COVID-19-related complications [50], and not selected according to a specific clinical presentation. They observed pathological features of recent ischaemia in six of the brains, which was most likely a result of reduced blood and oxygen supply to the brain, secondary to diffuse alveolar lung damage and multi-organ failure, and in keeping with the previous report from Solomon et al.[36]. More importantly, there was noticeable activation of microglial cells and a few perivascular T lymphocytes in all cases (Figure 1). It is likely that these findings are secondary to severe systemic infection with multi-organ failure and being on a ventilator, and this is further supported by the failure to detect viral RNA or protein in a subset of four cases examined by immunohistochemistry, in situ hybridization (ISH) and reverse transcriptase-PCR. However, one cannot exclude the possibility of low-grade viral infection below the sensitivity of the conducted techniques, particularly in those cases which show more than occasional perivascular T lymphocytes, or intense activation of microglial cells [55]. This view is supported by experimental work from Netland et al. who suggested that a minimal inflammatory cell process was observed in experimental mice infected with SARS-CoV even when infection was widespread, which raises the possibility of neurotoxicity associated with minimal cellular infiltration [14]. A previous report suggests a mechanism of endothelial cell infection and lymphocytic endotheliitis related to SARS-CoV-2 infection, which was associated with accumulation of lymphocytes around blood vessels as well as apoptotic bodies [15]. While electron microscopy was not performed in the authors’ cohort, the light microscopic examination did not detect changes in endothelial cells and the T lymphocytes were clearly in perivascular spaces with no infiltration into the blood vessel walls or endothelial cells.

One of the examined cases showed haemorrhagic infarction associated with mucormycosis and microthrombi (Figure 2). Mucormycosis is a relatively uncommon fungal infection suggesting that this individual had reduced immunity, possibly secondary to COVID-19. Lower lymphocyte counts in those with CNS symptoms with COVID-19 and immune deficiency pathogenesis has been reported. There were also microthrombi in small blood vessels around and within the infarcted brain tissue which could have developed secondary to the infarction and are consistent with the observation that cerebrovascular events are one of the common complications of COVID-19.

Additionally, within this study of eight cases, a case of encephalitis was reported, which is likely a more specific neuropathological feature of COVID-19. The brain showed intense and localized T lymphocyte infiltration and activation of microglial cells to the medulla, without necrosis, oedema, demyelination or meningitis (Figure 3). There was no inflammation in the rest of the sampled brain regions. Clinically, there was no evidence of encephalitis apart from dysphagia, which may be a non-specific indicator of brain stem involvement. The close proximity of the inflammatory foci to the reticular formation and cardiorespiratory centre in the medulla suggest that brain stem COVID-19 encephalitis might have disrupted the function of the cardiorespiratory centre. This case was one of four brains subjected to extensive investigation for SARS-CoV-2 infection by immunohistochemistry, ISH and RT-PCR, all of which were negative. However, as discussed above, the possibility of direct infection by SARS-CoV-2 cannot be completely excluded due to the potential lack of sensitivity in using formalin fixed paraffin-embedded tissue, other technical reasons, or even that the virus is no longer present in the tissue despite inflammation remaining. The identification of active and replicative/infective SARS-CoV-2 RNA in the brain is still not definite [36, 54, 55]. Indeed, in viral encephalitis involving the most prevalent viruses known to reach the brain (mainly herpes viruses, arboviruses and enteroviruses), the viral antigen can only be detected in 3–30 out of 100 000 infected persons [57]. The other differential diagnoses such as opportunistic (non SARS-CoV-2) viral infection (due to reduced immunity)[12] or autoimmune encephalitis, have also to be considered.

Future work

More detailed work involving a large number of autopsy brain examinations from COVID-19 patients, combined with close correlation with neurological manifestation, neuroimaging, general autopsy findings and isolation of SARS-CoV-2 from brain tissue and cerebrospinal fluid is required to clarify the role of CNS involvement in disease process and mortality.