Neurological symptoms in COVID-19: a cross-sectional monocentric study of hospitalized patients

In December 2019, a high number of patients infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was first reported in Wuhan, China and marked the beginning of a pandemic of unknown dimension in the twenty-first century. Although Corona Virus Disease 2019 (COVID-19) primarily affects the pulmonary system, neurological pathological manifestations in patients infected with SARS-CoV-2 have been reported already in early stages of the pandemic [1]. The growing number of studies on SARS-CoV-2-induced central nervous system (CNS) effects revealed a multitude of neurological implications ranging from meningitis, encephalitis, vasculitis, acute disseminated encephalomyelitis, neuropathies to SARS-CoV-2-associated strokes [2]. Several single-case reports of rapidly developed parkinsonism after SARS-CoV-2 infection gave rise to the idea that SARS-CoV-2 could also accelerate or aggravate pre-existing neurodegenerative disorders, or even induce them de novo [3, 4].

SARS-CoV-2, similar to the related corona viruses SARS-Coronavirus-1 (SARS-CoV-1) and the Middle East respiratory syndrome-related coronavirus (MERS), is potentially neurotrophic, and several mechanisms of CNS and peripheral nervous system (PNS) invasion have been proposed. The detection of viral particles and genomic sequences of SARS-CoV-1 in lymphocytes and monocytes implicates a hematogenous way as a possible mechanism of viral entry into the CNS [5, 6]. Another way of CNS invasion might be SARS-CoV-2 interaction with Angiotensin-converting enzyme 2 (ACE2) receptors which are not only expressed in lung and intestinal epithelium but are also found in the endothelial cells of the blood-brain barrier (BBB) [7]. Noteworthy, SARS-CoV-2 – like Influenza virus – can enter the brain via the olfactory nerve and induce an inflammatory status with subsequent immune reactions that affect surrounding brain regions as well [8]. In addition to the effects induced by the virus itself, other mechanisms including indirect effects of SARS-CoV-2 infection such as massive cytokine release, hypercoagulopathy and the organ damage it triggers, or sepsis itself have also been described as possible mechanisms responsible for sequelae [2].

One of the first local outbreaks of SARS-CoV-2 in Germany was registered in the neighboring districts of Heinsberg, leading to a high number of patients with COVID-19 transferred to the RWTH University Hospital Aachen. The neurological complications of COVID-19 patients that have been reported with increasing frequency are extremely diverse [9]. Therefore, we aimed a systematically investigation of neurological manifestations of COVID-19 patients treated at the RWTH University Hospital Aachen during the first pandemic wave. Our aim was to better understand the spectrum of neurological symptom manifestation in the context of their global occurrence.

We present data from a single-center prospective study of 53 in-house COVID-19 patients hospitalized at the RWTH University Hospital Aachen in Germany, which were referred to the Department of Neurology between March and September 2020. From March 2020 on, we tried to systematically examine all admitted COVID-19 patients. However, in-depth neurological characterization was only possible in 53 of these 138 patients. There were several reasons that made it impossible to recruit more patients: (i) Initially, COVID-19 was considered a disease mainly affecting the lung, but this view was challenged soon. (ii) As soon as patients were diagnosed with COVID-19, they were isolated in dedicated COVID-19 wards. Contacts were restricted as much as possible and particularly in the beginning of the patients’ stay, neurological consultation had not been deemed necessary. However, the frequent occurrence of neurological symptoms changed this view throughout the initial weeks. (iii) The University Hospital Aachen being reference center, COVID-19 patients were transferred but their critical status did not allow for in-depth neurological examination or neurological imaging. (iv) Of the patients treated in the Intensive Care Units (ICU), 44% died [10]. Patients included in this report were systematically examined for neurological impairment. Patients with proven SARS-CoV-2 infection were isolated in specialized general wards in the Departments of Cardiology and Pulmonology, respectively, as well as Intensive Care Units (ICU), and were examined by neurology consultants (UE or MR).

A confirmed case of COVID-19 was defined as a positive real-time reverse-transcriptase–polymerase-chain-reaction (RT-PCR) of sample material from either bronchoalveolar lavage of intubated patients (n = 28) or nasal swab (other patients, n = 25).

Patients were further categorized according to the presence of acute respiratory distress syndrome (ARDS). According to the Berlin definition, ARDS is classified as mild, moderate, or severe by the relation of arterial partial pressure of oxygen (PaO2) to fraction of inspired oxygen (FiO2) at a threshold of 300, 200, and 100 mmHg, respectively [11].

We collected data on medical history, comorbidities such as vascular risk factors, cardiological and neurological pre-existing conditions and medication, from the patients’ clinical records. To obtain a comprehensive picture of neurological comorbidities, a detailed history regarding central and peripheral neurological symptoms was further obtained for each patient.

Neurological examination of intubated and sedated patients included assessment of consciousness using the Glasgow coma scale (GCS), screening for orientation and delirium, neck stiffness, examination of cranial nerves by assessment of pupils for size, symmetry and reactivity to light, primary eye position, motor and sensible response, deep tendon reflexes and pathologic reflexes (Babinski sign). If during weaning phase, patients were able to follow commands, then motor (tone, signs of rigidity and spasticity), sensitivity and coordination tests were carried out. Patients treated on the general isolation wards received a more comprehensive neurological examination including mental status, cranial nerves, motor system, deep tendon reflexes, sensitivity, coordination testing, nystagmus, tremor, assessment of ataxia and gait. With alert, conscious and cooperative patients, we performed the Montreal Cognitive Assessment Test (MoCA) to screen for cognitive impairment [12].

If further diagnostic clarification was needed following the neurological examination, lumbar puncture and cerebrospinal fluid (CSF) analysis, cranial computed tomography scan (CT) or magnetic resonance imaging (MRI) scan and/or electroencephalography (EEG) were carried out. Written informed consent was waived upon of urgent medical indication, particularly for patients who underwent invasive mechanical ventilation on ICU. For conscious patients, informed consent was obtained (RWTH University Hospital Aachen’s ethics committee approval number: 148/20). CSF was examined for elevation of cell count, level of protein, glucose, lactate, blood-brain-barrier (BBB) function, oligoclonal bands and presence of SARS-CoV-2 RNA. In addition, routinely collected laboratory parameters were evaluated with particular interest on D-dimers, Interleukin-2 (Il-2), Interleukin-6 (Il-6), ferritin, tumor necrosis factor alpha (TNF-alpha) and neuron-specific enolase (NSE).

Structural and vascular brain imaging were conducted either on CT, 1.5-Tesla magnetic resonance imaging (MRI), or 3-Tesla MRI. For most cases T1-weighted spin-echo (partly with gadolinium-based contrast agent), diffusion-weighted imaging (DWI), apparent diffusion coefficient (ADC)-imaging, gradient-echo T2, T2* susceptibility-weighted imaging, 2D FLAIR, and Time-of-Flight-Angiography (TOF) were available.

Electroencephalography (EEG) was done using the Sigma Medizin-Technik/Neurowerk Center V10.0.0.10-system. EEGs were recorded over 20 min with scalp electrodes placed according to the International 10–20-system. Low-pass and high-pass filters were set to 70 Hz and 0.3 Hz, respectively. Intensive Care Unit (ICU)-patients were examined using an eleven electrode-system including frontal (F), temporal (T), parietal (P), occipital (O) and central (Cz) electrodes. Even-numbered electrodes refer to electrode placement on the right side of the head, whereas odd numbers refer to those on the left.

A “z” refers to an electrode placed on the midline sagittal plane of the skull, (Fz, Pz, Cz). F3, F4, T3, T4, P3, P4, O1, O2 and Cz. Mobile patients from general isolation wards were examined using further electrodes Fp1, Fp2 (pre-frontal), F3, F4, T1, T2, T3, T4, T5, T6, P3, P4, O1, O2, Fz, Pz and Cz. Patients were stimulated by verbal commands, eye opening, and if necessary, by sternal rub (For a chronological list of all investigated patients with COVID-19 see Additional file 1).

Continuous variables were expressed as either medians and interquartile ranges or as simple ranges as adequate. Categorical variables were summarized as frequencies and percentages, respectively. No imputation was made for missing data. Since the sample of patients in our study was not derived from random selection, all statistics are descriptive only. Statistical analyses were performed using SPSS IBM version 26 with a type-I-error (α) set to 0.05 as the statistical threshold for significance. Groups were compared using non-parametric approaches using Pearson Chi², Fisher’s Exact test, or Mann-Whitney U-test, depending on the comparison, as indicated.

In the present study, we systematically investigated neurological deficits in COVID-19 patients. While numerous publications are available on clinical manifestations of SARS-CoV-2 infections by now, some of them focusing on neurological aspects mostly in the form of case reports or small case series, no systematic neurological investigation of in-house patients on both normal wards and intensive care units are available from Germany so far.

COVID-19 patients hospitalized at the RWTH Aachen University Hospital during the first wave of the pandemic (March through September 2020) presented mainly with the following neurological features: Coordination deficits (74%), cognitive impairment (61.5%), paresis (47%), abnormal reflex status (45%), sensory abnormalities (45%), general muscle weakness and pain (32%), hyposmia (26%), and headache (21%). Among them, patients with ARDS were neurologically more severely affected than non-ADRS patients.

Whereas 29.6% of our patients presented with sulcal bleedings, only 11.1% of our patients showed ischemic stroke associated with SARS-CoV-2 infection. Numbers of COVID-19-associated cerebrovascular events vary in the literature, with a single to high number of cases reported by different observational studies, e.g. 5% ischemic stroke in Wuhan, China [13], 2% in Milan, Italy [14], and in the Netherlands [15], while other colleagues from Brescia in Italy reported 77% [16] or 45% in the U.K. [17]. The etiology of the strokes remains unclear. While some researchers suggested a direct effect of the viral infection by either positive SARS-CoV-2 PCR in CSF or specific intrathecal antibody synthesis [2], many of the stroke-COVID-19 patients already had traditional cardiovascular risk factors. Since those patients with pre-existing cardiovascular risk factors and disease have a higher risk not only for stroke but also for a severe course of COVID-19, it is difficult to pinpoint the actual stroke etiology. What complicates things is the fact that many ICU patients suffer from disseminated intravascular coagulation [18], what further increases the risk of thromboembolic pathologies. Elevated D-dimer, fibrinogen, and the presence of antiphospholipid antibodies appear to be prominent in COVID-19 patients with concomitant acute ischemic stroke [19]. Most likely, stroke is rather correlated with COVID-19 and its hematological complications than causally are linked to the viral replication itself.

In contrast to others [2], in our cohort we did not observe a single case of SARS-CoV-2-related meningitis [6] nor acute disseminated encephalomyelitis myelitis [20‐22], neither did we observe Guillain-Barré syndrome [23, 24]. Roughly 24% of our patients were diagnosed with critical illness neuro−/myopathy (CIN/CIM), a percentage that is similar to neurological ICU patients treated for other indications. Interestingly, 26% patients reported hyposmia, which is generally in line with previous reports of about 20% hyposmia in COVID-19 patients [25], emphasizing the neuro-invasive potential of SARS-CoV-2. Our EEG findings were rather unspecific. Pathological EEG findings were reported in form of epileptiform discharges and frontal sharp waves in COVID-19 patients with impaired consciousness or suspected clinical seizures [26] and most likely reflect diffuse encephalopathy.

Among the nine patients with lumbar puncture, none was positive for SARS-CoV-2 RNA in the CSF. There are single reports of positive RT-PCR results from CSF [27, 28], whereas in the majority of cases the PCR remains negative in spite of the patient’s neurological symptoms, even in cases of encephalitis [29‐36]. There are various hypothesis about the viral invasion ways into the CNS, such as hematogenous spread by introduction of the virus via infected peripheral leukocytes, ACE2-mediated transmission through endothelial cells, or trans-synaptic spread into the brain via the olfactory epithelium in the nasal cavity [5]. The latter seems to be likely due to the spatial proximity of the CNS to the pharynx with high viral load. An alternative way is hematogenous spread via virus-associated damage to the BBB. In most patients with e.g. bacterial meningitis, the pathogen reaches the brain via hematogenous ways and not by propagated infection of the throat or ear, which makes hematogenous spreading of SARS-CoV2 into the CNS a possibility.

In ten patients (18.9%) the cranial nerve was affected with SARS-CoV-2, symptoms including oculomotor and/or facial nerve palsy and/or trigeminal nerve deficits. Out of these ten patients, three showed multiple cranial nerve involvement. Of the ten patients with cranial nerve involvement only two received a lumbar puncture for CSF analysis, which revealed mild pleocytosis and elevated protein concentrations in both cases (Table 4, cases #2 and #32). In both cases we detected reactivation of other viruses, such as Epstein-Barr-virus and Herpes-virus. One patient was treated with Ganciclovir, the other with Aciclovir. In summary, the diagnosis of neuritis cranialis is likely, because (i) SARS-CoV-2 PCR from CSF was negative in all our cases with CSF analysis, and (ii) total CSF protein was increased in these two cases with cranial nerve involvement but not in the majority of the other cases without cranial nerve involvement (Table 4).

In general, cerebral complications are common in patients with veno-arterial extracorporeal membrane oxygenation (ECMO) independent of Covid-19. ECMO therapy is associated with an increased risk for ischemic lesions as well as hemorrhagic lesions. In our cohort, four of the eight patients with proven subarachnoid hemorrhage were treated with ECMO. All patients in our cohort with subarachnoid hemorrhage received anticoagulation for therapeutic purpose. For instance, one patient who was not treated with ECMO had a myocardial infarction, another had thrombosis of the subclavian and jugular veins. Since the subarachnoid hemorrhages were not typical of aneurysm hemorrhage and we considered other possible causes more likely, no angiography was performed.

In our cohort, the number of patients with cognitive impairment was astonishingly high with 61.5% and must be interpreted with caution since these observations were acquired during the acute phase of COVID-19. MoCA is a good bedside test to detect cognitive decline but can yield aberrant results when applied to patients with dementia, neurodegenerative diseases of different origin, or in patients with septic encephalopathy. The degree of cognitive pathology in our cohort is high in comparison to roughly 6% (10/153) of patients presenting with “dementia-like” syndrome in the UK [17]. One of the cognitively impaired patients had a pre-existing diagnosis of a drug-treated idiopathic Parkinson’s disease, but in the other patients no pre-existing cognitive impairments were known. The other pre-existing medical conditions included arterial hypertension, diabetes mellitus to hepatitis C, polysubstance use, non-small-cell bronchial carcinoma, and metastatic cholangiocellular carcinoma.

Since the cognitive impairment of our patients during the acute phase of COVID-19 constitutes only a snapshot of the situation, further longitudinal assessment is needed to determine potential long-term complications and co-morbidities. Our data are not without a selection bias. We were only able to neurologically characterize 53 of 138 patients in depth (38.4%), although we tried to neurologically examine all patients beginning end of March 2020, regardless of the severity of the Covid-19 disease. Because patients were primarily in internal medicine wards rather than neurology wards, we missed out on some patients who, for example, died shortly after admission. The initial assumption that Covid-19 was only a pulmonary disease, was also partly responsible for the loss of patients at the beginning of the pandemic. However, because we examined all patients regardless of the severity of the disease, our report reflects the distribution of neurological symptoms whose disease stage required admission to a university hospital. Taken together, neurological abnormalities in the COVID-19 patient cohort assessed in our study were not uncommon, yet not very specific. The neurological complications of SARS-CoV2 are similar to those of other coronavirus’ diseases such as SARS and MERS [37]. It is not surprising that the most severely affected patients with ARDS or on ICU had more neurological deficits and complications than the non-ARDS patients. Some of the observed neurological symptoms are commonly found in the context of severe infectious diseases and after ICU interventions in patients with e.g. septic encephalopathy, and are therefore no unique features of COVID-19. On the other hand, other symptoms such as hyposmia, which was common in our patients like in various other studies [38‐43], seem to be comparably SARS-CoV2-specific and may indeed represent one possible pathomechanism of the virus penetration into the CNS [5]. The quite common occurrence of hemorrhage with a predilection for the corpus callosum can be confirmed in out cohort. Microbleeds of the splenium are seen following traumatic head injury or in critical ill patients with a need for mechanical ventilation [44] and are - like in our patients - associated with leukencephalopathy. Both findings were predominant in male patients who showed stronger severity of symptoms and who had been hospitalized for a longer time, with worse functional outcome at discharge, or even death [45]. We can confirm this clientele in our cohort with abnormal MRI findings in patients with ARDS and/or prolonged ventilation.

In the majority of data it remains difficult to identify specific (e.g. neurotropism of SARS-CoV2, possible intrathecal viral replication) and non-specific neurological features (e.g. hypoxia, septic encephalopathy, CIN/CIM) of a disease primarily affecting the respiratory tract. Taken together, we interpret our neurological observations as mostly associated with hypoxia and septic states induced by a strongly pro-inflammatory viral infection and to a far lesser extend caused by direct viral neuronal damage.

We are aware of the limitations regarding the interpretation of cognitive deficits found in this cohort, particularly in the context of the acute illness and especially in febrile or delirious patients. We therefore interpret these data rather cautiously, also because the premorbid cognitive state of the patients is unknown. Nevertheless, we assume that previously cognitively impaired patients can experience a clear and permanent aggravation in cognition after acute illnesses and that even previously cognitively non-impaired patients can manifest with dementia after severe infectious diseases associated with sepsis or delirium [46]. Since our data is limited in the context of acute events, a follow-up after 3 and 9 months, respectively, including neuropsychological testing and MRI is currently underway to detect possible permanent sequelae after COVID-19, allowing us to assess possible long-term deficits.