Central Nervous System Disorders in Severe SARS-CoV-2 Infection: detailed clinical work-up of eight cases

Emanuela Keller (  emanuela.keller@usz.ch ) Department of Neurosurgery and Institute of Intensive Care Medicine, University Hospital and University of Zurich, Switzerland https://orcid.org/0000-0002-7560-7574 Giovanna Brandi Institute of Intensive Care Medicine, University Hospital and University of Zurich, Switzerland Sebastian Winklhofer Department of Neuroradiology, University Hospital and University of Zurich, Switzerland Lukas Imbach Department of Neurology, University Hospital and University of Zurich, Switzerland Daniel Kirschenbaum Department of Neuropathology, University Hospital and University of Zurich, Switzerland Karl Joachim Frontzek Department of Neuropathology, University Hospital and University of Zurich, Switzerland Peter Steiger Institute of Intensive Care Medicine, University Hospital and University of Zurich, Switzerland Sabeth Aurelia Dietler Institute of Intensive Care Medicine, University Hospital and University of Zurich, Switzerland Marcellina Isabelle Haeberlin Department of Neurology, University Hospital and University of Zurich, Switzerland Jan Folkard Willms Institute of Intensive Care Medicine, University Hospital and University of Zurich, Switzerland Francesca Porta Intensive Care Unit, Graubuenden Cantonal Hospital, Chur, Switzerland Adrian Waeckerlin Intensive Care Unit, Graubuenden Cantonal Hospital, Chur, Switzerland Irene Alma Abela Department of Infectious Diseases and Hospital Epidemiology, University Hospital and University of Zurich, Switzerland Andreas Lutterotti Neuroimmunology and Multiple Sclerosis Research Section, Department of Neurology, University Hospital and University of Zurich, SWitzerland Christoph Stippich Department of Neuroradiology, University Hospital and University of Zurich, Switzerland Ilijas Jelcic Neuroimmunology and Multiple Sclerosis Research Section, Department of Neurology, University Hospital and University of Zurich, Switzerland

Comprehensive workups of as many cases as possible are needed to understand a potential "Neuro-COVID-19" disease and, in a next step, to develop preventive as well as therapeutic strategies.
Below eight cases of critically ill patients with the primary diagnosis of severe COVID-19 pneumonia and subsequent neurologic complications examined with electroencephalography (EEG), neuroimaging and cerebrospinal uid (CSF) analysis are presented. The ndings are assigned to previously published case reports.

Study Design and Participants
We retrospectively analysed consecutive patients with severe COVID-19 treated at the Institute of Intensive Care Medicine, University Hospital Zurich from March 09, 2020 to April 03, 2020. The patients were evaluated for concomitant severe central nervous system (CNS) symptoms occurring during their critical disease state. Furthermore, the Institute of Intensive Care Medicine of the University Hospital Zurich supported the Graubuenden Cantonal Hospital by the means of consultative treatment regarding an additional critically ill COVID-19 patient with CNS symptoms who had been admitted there in the same time period. This case is also included in the study. The study was performed in accordance with the principles of the Declaration of Helsinki and was approved by the local ethics committee (ID 2020-00894). Written consent was given by the legal representatives as all patients were still incapable of judgment at hospital discharge.

Data Collection
Severe cerebral disorders triggering the additional examinations such as cranial computed tomography (CT) or magnetic resonance imaging (MRI), EEG as well as CSF analysis were: impaired consciousness (negative or delayed wake-up attempt after termination of analgosedation with persistent coma, stupor or delirium), acute cerebrovascular disease, clinical seizure / status epilepticus and myoclonic movements. At time of neurological assessment, midazolam serum levels were below the detection limit in all patients. SARS-CoV-2 infection was con rmed by reverse-transcription polymerase chain reaction assay (RT-PCR) in throat and or tracheobronchial samples (according to (12) or by Roche Cobas SARS-CoV-2 test). Metagenomic virus sequencing was performed as described (13). All examinations (laboratory values, chest radiographs and CTs) as well as neuroimaging, EEG and CSF analysis were performed according to the clinical needs of the patients. Electronic medical records were reviewed for date of COVID-19 symptom onset, neurological symptom onset, age, sex, pre-existing medical conditions (arterial hypertension, diabetes, smoking, cardiac disease, cerebrovascular disease, immunosuppression because of malignancy or immunosuppressive treatment, lung disease, treatment with ACE inhibitors or angiotensin II receptor antagonists, statins, anticoagulants) and speci c initial symptoms, especially anosmia/hyposmia, neuropsychiatric symptoms and focal neurologic de cits. Data on empiric COVID-19 therapies (Hydroxychloroquine, Remdesivir) and pre-intubation worst oxygen saturation (SpO 2 ) were recorded. In addition to daily routine laboratory tests (blood cell counts, chemistry, coagulation testing) in patients with neurological de cits, serum was analysed for glucose, lactate, immunoglobulins indices, interleukin-6, antinuclear antibodies (ANA), anti-neutrophil cytoplasmic antibodies (ANCA), antibody to native DNA (nDNA), lupus anticoagulant, anti-cardiolipin, anti-β2-glycoprotein I, anti-SSA/Ro, anti-SSB/La and anti-IgLON5 antibodies.
Patient serum was tested for antibodies for viral infections (including SARS-CoV-2, herpes simplex virus (HSV), varizella zoster virus (VZV), cytomegaly virus (CMV), hepatitis B virus (HBV) and human immunode ciency virus (HIV)). Laboratory values on the day of CSF sampling or the day before were analysed. CSF samples were analysed for cell count (white and red cells) and cell composition, chemistry (glucose, lactate and protein), bacterial culture, neurotropic viruses using RT-PCR to detect HSV, VZV, CMV and SARS-CoV-2 and immunology parameters including oligoclonal bands, total CSF/serum IgG-, IgA-and IgM quotients as well as SARS-CoV-2-, HSV1-, CMV-, VZV-speci c and anti-IgLON5 antibodies. SARS-CoV-2 antibodies in serum and CSF were tested using an in-house developed, bead-based antibody assay using Luminex that detects IgG, IgA and IgM antibodies against subunit 1 (S1) and subunit 2 (S2) of the spike protein and nucleoprotein (NP) of SARS-CoV-2. Mean uorescence intensity (MFI) was assessed using FLEXMAP 3D (Luminex). Antibody reactivity was assessed in arbitrary units (AU) using a standard curve obtained from a serial dilution of the respective standard serum for interpolating MFIs by four parameter logistic curve t. AUs within the linear part of the standard curve were multiplied by the corresponding dilution factor to obtain absolute AU. To calculate intrathecal SARS-CoV-2-, HSV1-, CMV-, and VZV-speci c IgG synthesis, the antigen-speci c CSF/serum antibody index (CAI spec ) was calculated according to (14).
We recorded 20 minutes EEGs using standard 25 channel montage (10/20). Electrophysiological analysis included visual scoring for epileptic activity, vigilance, background activity, focal slowing and quantitative spectral analysis. We performed standardized painful and acoustic stimuli to test for EEG reactivity in all patients.
Intracranial MRI vessel wall imaging was performed in three patients using high resolution T1w dark blood non-contrast and contrast enhanced space sequences (Skyra, Siemens Healthineers, Forchheim, Germany).

Statistical Analysis
Descriptive statistics and frequency analysis are calculated to describe characteristics of the study sample and overall prevalence of concomitant factors. Mean and standard deviation (SD) are used for normally distributed data and median and range for data that were not normally distributed.
Categorical variables are expressed as counts and percentages. The time from onset of the rst symptoms of COVID-19 to manifestation of CNS symptoms was calculated in days.

Demographic and Clinical Characteristics
Among 32 critically ill patients with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection eight patients (18%) presented with severe CNS involvement. Their mean (SD) age was 67.6 (6.8) years, seven were male. Two patients died due to respiratory or multiorgan failure. Baseline characteristics are given in Table 1. All patients had cardiovascular risk factors, most frequently hypertension as a pre-existing condition. The characteristics of the patients during the course of the disease are given in Table 2. All patients were treated with invasive mechanical ventilation and all but one with prone positioning. Seven had to be treated with continuous renal replacement therapy due to renal failure. Three patients suffered from myocardial injury characterized by elevated troponin, myoglobin and creatinine phosphokinase (patients 1,2 and 6) and one from acute mesenteric ischemia requiring surgery and leading to multiorgan failure (patient 1).

Clinical and Neuroradiological Characteristics
Page 4/12 CNS symptoms became manifest on day 17.5 in median (range day 2 to 34). In no patient anosmia/hyposmia or neuropsychiatric symptoms could be revealed in the medical history. In two patients (patients 6 and 7), CNS symptoms included primary focal neurological de cits due to an ischemic stroke, in one of them manifesting before pneumonia became evident (patient 7). In both patients single lacunar ischemic lesions corresponding to their primary focal neurological de cits were present in the initial CT. Their follow-up MRI scans showed additional small ischemic lesions in different vascular territories. All but one patient (patients 1, 2, 4, 5, 6, 7 and 8) presented with impaired consciousness for several days after termination of analgosedation. In all but two patients (patients 3 and 4), CT or MRI scans performed later in the course of the disease, showed multiple cerebral microbleeds, in three of them with additional subarachnoid haemorrhage (patients 1, 2 and 8) (Figure 1). Three patients were examined with intracranial vessel wall sequence MRI (patients 2, 6 and 7). In all three patients vessel wall contrast enhancement of large to middle-sized cerebral arteries was found ( Figure 2).
One patient was admitted from a peripheral hospital due to status epilepticus (patient 3) which was successfully treated with levetiracetam and valproate. Thereupon pronounced myoclonic movement patterns occurred on all extremities. Except generalized brain atrophy, CT and MRI scans showed no abnormal ndings.

Laboratory Findings
Laboratory ndings are given in table 3

EEG Findings
All patients showed similar, yet unspeci c EEG patterns with severe to moderate background slowing and anterior rhythmic delta activity consistent with metabolic encephalopathy. No interictal epileptiform discharges were found. However, we found stimulus induced periodic discharges (patient 5) and rhythmic frontal theta activity in myoclonic status (patient 2) as a correlate for enhanced epileptogenicity.

Autopsy Findings
Brain autopsy of one patient (patient 1) showed acute microbleeds in the pontine tegmentum and acute microinfarcts in the basal ganglia. Around the rostral surface of the cerebellum an extensive, acute subarachnoidal haemorrhage was seen. Adjacent cerebral tissue showed multiple fresh, partially con uent, microinfarcts and parenchymal haemorrhages. In addition, fresh blood was found in the perihippocampal area.

Outcome
At time of submission all six surviving patients were still hospitalized but showed signi cant neurological recovery (three with still mild cognitive de cits, two with slight to moderate tetraparesis indicating critical illness neuro-and myopathy, one with persistent myoclonic movements). The two stroke patients showed no focal neurological de cits anymore.
In the present patient population of 32 ICU patients with COVID-19, eight (25%) had severe CNS symptoms, among them two with hemiparesis and lacunar ischemic stroke and one with status epilepticus in the early phase. As most common presentation, six patients presented with prolonged impaired consciousness after termination of analgosedation, three with temporarily delirious state.
The different patterns of our cases suggest that variable pathogenesis may lead to CNS disorders in COVID-19. Evaluating and summarizing the published small series and individual cases (1-3, 5-11), neurologic sequelae can be assigned to the following pathophysiological groups: Unspeci c, related to severity of disease In one case hypoxic encephalopathy, presenting as status epilepticus and myoclonic movements, and in another case, metabolic-toxic encephalopathy, associated with uraemia, might be unspeci c cerebral complications in two of our critically ill COVID-19 patients. In both patients neuroimaging showed no speci c ndings but frontotemporal brain atrophy, which might be associated with sepsis-induced brain dysfunction (15). Sepsisassociated encephalopathy, which typically presents as confusion and coma, is reported in up to 70% of patients with sepsis (16). All our patients' EEG examinations showed severe generalized background slowing, indicating global cognitive dysfunction and occurring in patients after recovery from severe sepsis (17). In the Beijing series, critically ill COVID-19 patients developing septic shock and multiorgan failure have been described (18). In 76% of the cases cultures for bacteria and fungus were negative. The authors, therefore, used the term "viral sepsis". As in sepsis induced by other microorganisms, hyperactivation of proin ammatory cytokines and chemokines play an important role in severe COVID-19 and may be associated with septic encephalopathy. Finally, epileptic activity seems to occur only rarely in COVID-19 patients (0.5% in the Wuhan series) (1) and might be related to hypoxia, multiorgan failure, and metabolic disorder rather than to CNS infection (19). In agreement, also in our series the occurrence of epileptic activity and myoclonic status in patient 3 was most probably caused by hypoxic brain injury.
Large and small cerebral vessel disease Two of our patients presented with acute ischemic stroke in the early phase of their disease. A series of ve cases of large -vessel stroke in patients younger than 50 years has been described in New York City (6). In the Wuhan series, the incidence of stroke in hospitalized patients was about 5% (1). Severe infections, especially respiratory-related, are known to trigger acute cerebrovascular events (20). Severe COVID-19 mostly develops in patients with cardiovascular risk factors and is characterized by a pronounced proin ammatory early phase (18), both factors even aggravating the risk for stroke at disease onset. In Strasbourg, 23 patients with neurological symptoms were examined with MR imaging (4). Three asymptomatic patients had small ischemic strokes and bilateral frontotemporal hypoperfusion was noted in all 11 patients who underwent perfusion imaging (4). Contrastenhanced perfusion MR was not performed in our series. However, the unspeci c EEG patterns in our patients with background slowing and occasional anterior rhythmic delta activity, consistent with unspeci c encephalopathy, might also re ect frontal hypoperfusion.
Further extraordinary ndings in our series were detected later in the course of the disease. In all but one patient with delayed wake-up, neuroimaging or autopsy showed multiple cerebral microbleeds, in three of them with additional subarachnoid haemorrhage and in another two with additional small ischemic lesions, indicating CNS small vessel disease. Different factors might explain the involvement of small cerebral vessels. All patients had cardiovascular risk factors, most of them hypertension. The distribution of the microbleeds, however, was typical for hypertension in only one patient. None of our patients had a history of cerebral amyloid angiopathy. As in other series, all our patients had a hypercoagulable state with increased brinogen and D-dimer levels (11,21), which might lead to thrombosis of small cerebral vessels. Only one patient, however, ful lled the criteria of disseminated intravascular coagulation disorder (> 5 points according to the International Society on Thrombosis and Haemostasis criteria) (22), which is known to be associated with acute cerebral micro-angiopathy. Immune-mediated and infectious vasculitis may cause CNS small vessel disease (23) and a certain type of vasculitis involving cerebral vessels might be induced by COVID-19 as well.
In three patients with severe COVID-19 intracranial vessel wall sequence MRI scans were performed, for the rst time to our knowledge. MR vessel wall imaging showed contrast -enhancement of vessel walls in large and middle-sized cerebral arteries, suggesting vascular wall pathologies with an in ammatory component. However, MRI vessel wall contrast-enhancement is not speci c for vasculitis involving cerebral vessels. None of our patients showed in ammatory signs in the CSF or characteristic autoantibodies indicating systemic vasculitis. Chen et al. found vessel wall contrastenhancement in 45.8% of the patients with reversible cerebral vasoconstriction syndrome (RCVS) (24). The authors suggest that, among other factors, endothelial dysfunction might contribute to vascular wall in ammation in RCVS. However, none of our patients had typical signs of RCVS such as thunderclap headache and/or reversible multifocal cerebral vasoconstrictions.
Contrast -enhancement of vessel walls as well as the pattern of micro-bleedings and multiple small infarctions indicate that large, middle-sized as well as small cerebral vessels are involved. An autopsy study in patients with severe COVID-19 revealed vascular involvement in multiple organs (25). Lymphocytic endotheliitis was found in lung, heart, kidney, liver and small intestine. Furthermore, viral inclusion structures, could be found in endothelial cells. The angiotensin-converting enzyme 2 (ACE2), as the main host cell receptor of SARS-CoV-2 (26), albeit at lower concentrations, is also expressed in the endothelium and vascular smooth muscle cells of the brain (27). It might be hypothesized that the endothelium of brain vessels might be directly affected by the virus or that the virus induces a parainfectious immune-mediated in ammation of the endothelium. However, we did not nd any in ammatory CSF syndrome as a sign of infectious cerebral vasculitis in our patients or signs of perivascular in ammatory cell in ltration in the one post-mortem analysis performed. Another mechanism might be, that the cerebral vessels might be affected by the in ammatory state in the peripheral blood. Prolonged increased levels of IL-6, IL-10, IL-2 and IFN γ have been described in severe cases and may play an important role in the immunopathology of COVID-19 (28). In our patients, serum IL-6 was elevated in all but one patient up to levels of 594 ng/L. However, in all of them IL-6 values in the serum were higher than in CSF, which contradicts its intrathecal synthesis. Still, these observations do not exclude the possibility of damage to cerebral vessels induced by the hyperin ammatory state in the peripheral blood. Cytokines or in ammation-induced metabolic changes leaking from peripheral blood to the CNS micromilieu might lead to disseminated focal disturbances of the blood brain barrier and dysfunction of the respective surrounding brain tissue.

Encephalitis
Singles cases with COVID-19 and meningo-encephalitis, one case with acute necrotizing encephalopathy and one with acute disseminated encephalomyelitis (ADEM) have been described (2,3,5,(7)(8)(9)(10)(11). Neuroinvasion is hypothesized to occur via hematogenous or neuronal route as over the olfactory nerve (29). In two cases with neuropsychological symptoms, one of them with status epilepticus, lymphocytic pleocytosis but negative RT-PCR for SARS-CoV-2 was found in the CSF (2). In another case suffering from disorientation and hallucinations, CSF analysis also revealed lymphocytosis (3). To our knowledge, up to date, gene sequencing con rmed the presence of SARS-CoV-2 in the CSF in Beijing in only one case (30) and RT-PCR was positive in only one patient in Japan (5). None of our patients, not even those with status epilepticus, showed in ammatory signs neither in neuroimaging nor in the CSF. RT-PCR for SARS-CoV-2 in the CSF was negative in all patients. Intrathecal IgG production against the SARS-CoV-2 could not be demonstrated and even metagenomic virus sequencing was negative in two patients. Therefore, neither direct neuroinvasive potential nor directly CNS-directed parainfectious injury could be demonstrated in our patients.

Limitations
Only eight patients were studied. All were severely ill, entering the ICU with respiratory failure. Mild neurological symptoms at disease onset as headache, dizziness and taste or smell impairment could no longer be reliably evaluated. Furthermore, CNS symptoms might have remained undetected in patients who were under analgosedation. Diagnostics with neuroimaging and CSF analysis might not have been performed at the time of the occurrence of cerebral complications. Therefore, in ammatory signs in the CSF, might have been missed. Furthermore, RT-PCR tests for SARS-CoV-2 in the CSF are not yet su ciently clinically validated regarding sensitivity and timing of lumbar puncture after onset of symptoms. We cannot exclude the possibilities of (i) transient SARS-CoV-2 RT-PCR positivity in the CSF through true invasion of CNS with SARS-CoV-2, which turned out to be negative at the time of lumbar puncture, or (ii) concentrations of SARS-CoV-2 in the CSF below detection limit of our RT-PCR assay. Additional search for intrathecal SARS-CoV-2-speci c IgG production as a sign of humoral immune reaction against viral infection of CSF space turned out negative. In other viral CNS infections such as herpes encephalitis or tick-borne encephalitis, intrathecal antiviral IgG production usually consistently occurs 10-14 days after onset of symptoms, but may occur occasionally as early as 3 days after disease onset (31-33). Our observation of no intrathecal SARS-CoV-2 IgG production may be explained by the possibilities, that SARS-CoV-2 never entered CNS and therefore no intrathecal immune response was mounted or because intrathecal SARS-CoV-2-speci c IgG synthesis occurred after lumbar puncture. In analogy to other human viral CNS infections, the consistent combination of negative SARS-CoV-2 RT-PCR and absence of intrathecal SARS-CoV-2 IgG production 2 days (range 0-31) after onset of CNS disorder argues relatively against a direct invasion of CNS by SARS-CoV-2, but we cannot exclude the possibility of a differential diagnostic window, which might have been missed.

Conclusions
Severe CNS disorders are common in patients with severe COVID-19 and different pathomechanisms might be involved. Besides unspeci c encephalopathic patterns and encephalitic syndromes described as case reports, large vessel strokes might occur more often early after disease onset. In a later phase, microbleeds and microinfarctions as well as vessel wall contrast enhancement occur, indicating large and small cerebral vessels to be involved. Patients with endothelial dysfunction due to cardiovascular risk factors, mostly hypertension, might be especially at risk. Whether SARS-CoV-2 directly affects endothelial cells of brain vessels has to be investigated in basic research and more clinical cases. CNS disorders associated with COVID-19 can lead to long-term disabilities and may aggravate socio-economic damage. The mechanisms have to be investigated urgently in order to develop preventive and therapeutic neuroprotective strategies. 31. Fomsgaard A, Kirkby N, Jensen IP, Vestergaard BF. Routine diagnosis of herpes simplex virus (HSV) encephalitis by an internal DNA controlled HSV PCR and an IgG-capture assay for intrathecal synthesis of HSV antibodies.