1. Introduction
Coronavirus disease 2019 (COVID-19) is a multisystem viral sepsis syndrome that can affect different organ systems with symptoms ranging from mild to life threatening [
1]. Neurologic complications are commonly described and may occur as direct or indirect consequences of the viral infection, the treatment, the systemic inflammation due to immune activation or hypoxia or, in some cases, may be incidental associations [
2,
3,
4,
5]. However, the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is known to have neuroinvasive potential [
6,
7].
The accumulating data describing various neurological manifestations in COVID-19 are not only related to the acute phase [
3,
8,
9,
10] but are often part of a syndrome known as post-acute sequelae of COVID-19 (PASC) or long-COVID, with symptoms occurring and/or persisting for weeks or months after the initial infection [
8,
11,
12,
13,
14]. To date, whether the neurological manifestations are also accompanied by increased biomarkers of neuronal and astrocytic damage is still being investigated.
Brain injury biomarkers of neurological diseases have been investigated in the context of COVID-19. Indeed, during the acute stage of COVID-19, both cerebrospinal fluid (CSF) [
2,
15,
16] and plasma studies [
2,
17,
18,
19,
20] showed higher levels of neurofilament light chain (NfL) in patients with neurological symptoms (NS) which are correlated with disease activity, thus further supporting the occurrence of concomitant acute axonal injury. NfL is a subunit of neurofilaments, which are cylindrical proteins exclusively located in the neuronal axons, that can be measured in both CSF and plasma samples as a biomarker of neuronal injury [
21,
22]. Likewise, in the acute phase of COVID-19, it has been shown that increased levels of glial fibrillary acidic protein (GFAP) correlate with COVID-19 severity [
18,
23,
24,
25]. GFAP is an intermediate filament highly expressed in astrocytes regulating their morphology and function in the CNS [
26,
27]. Plasma GFAP levels are very low in healthy individuals, but increased GFAP levels due to astrocyte disintegration are known to indicate astrocyte damage [
26,
28]. Therefore, GFAP is increasingly used as a plasma biomarker of astroglial activation/injury [
21,
28].
Alteration of the blood–brain barrier (BBB) integrity has an important role in neurological diseases, including brain infections [
29]. Matrix metalloproteinases (MMPs), a family of enzymes that proteolytically degrade various components of the extracellular matrix (ECM), can have a detrimental effect, contributing to perturbation of the BBB integrity and neuroinflammation [
30]. Within the ECM, the tissue inhibitors of MMPs (TIMPs) inhibit and regulate MMP proteolytic activity. Among MMPs, MMP-9 is the most prominent in promoting BBB disruption associated with CNS damage and inflammation [
31]. In the acute stage of COVID-19, CSF levels of MMP-9 were found to increase and related to different inflammatory cytokines [
32] as well as to COVID-19 severity [
2].
Also, neurological diseases are associated with inflammatory conditions and several differences in the levels of biomarkers in neurological diseases compared to normal conditions were reported [
33]. Among these biomarkers, soluble CD163 (sCD163) levels increase in both CSF and plasma samples, related to the activation of macrophage/microglia and inflammation levels in neurological diseases, supporting its role also as a cognitive impairment biomarker [
34,
35,
36,
37]. sCD163 is a soluble form of CD163, a receptor for haptoglobin–hemoglobin complexes and a monocyte/macrophage-specific membrane protein. CD163 is found on macrophages in the CNS, including perivascular macrophages and microglia [
38] and its soluble form can be found in both CSF and plasma [
39] after monocyte/macrophage activation [
40,
41]. In the CNS, sCD163 is probably shed by macrophages and microglia, triggered by complex immune-modulating mechanisms in the microenvironment [
34]. Although plasma levels of sCD163 were deeply investigated in the acute stage of COVID-19 and correlate to COVID-19 severity [
42,
43,
44], data on CSF levels of sCD163 are lacking.
In the present study, in a cohort of patients with different COVID-19 severity, we longitudinally investigated the levels of brain injury (NfL and GFAP), myeloid activation marker (sCD163) and in turn, implications for the CNS during the acute stage of the disease and three months after hospital discharge. Moreover, in those patients requiring puncture lumbar (PL) for diagnostic purposes, the impairment of blood–brain barrier (BBB) (MMP-9 and TIMP-1) was also investigated.
2. Materials and Methods
2.1. Study Design
A single-center retrospective study involving hospitalized COVID-19 patients from March 2020 to March 2021 was performed. Specifically, at the Department of Public Health and Infectious Diseases, Policlinico Umberto I, Sapienza, University of Rome, (baseline) adult patients (≥18 years old) were enrolled upon hospital admission. As previously described [
45,
46], COVID-19-related pneumonia was diagnosed by high-resolution chest computed tomography (CT) scan associated with SARS-CoV-2 RNA detection from a nasopharyngeal swab through a commercial reverse transcription-polymerase chain reaction (RT-PCR) kit, following manufacturer’s instructions (RealStar
® SARS-CoV-2 Altona Diagnostic, Hamburg, Germany).
Three months after hospital discharge (Tpost), COVID-19 patients were followed up with at the post-COVID clinic of Policlinico Umberto I, Sapienza, University of Rome. During the post-COVID visit, each subject underwent a detailed interview with an infectious disease physician who asked questions related to post-COVID symptoms. Specifically, patients were queried regarding the presence of post-COVID symptoms, including NS such as trouble concentrating or with memory, headache, trouble with taste or smell.
At first, according to COVID-19 severity at the acute stage of the disease (based on acute respiratory distress syndrome [ARDS] onset), patients were stratified into two groups: severe and non-severe. Then, these two groups were further classified into four subgroups based on the maximum oxygen supply/ventilation support required during hospitalization. Specifically, the severe group was stratified into invasive mechanical ventilation via orotracheal intubation (IOT) and noninvasive ventilation (NIV) subgroups while the non-severe group was stratified into Venturi mask for oxygen (VMK) and room air (AA) subgroups.
Next, an additional stratification was performed based on NS presence at both time points into groups with and without NS. Specifically, at baseline NS presence was defined according to the necessity of PL for diagnostic purposes, while at Tpost, NS was self-referred by the patients during the post-COVID visit after being asked by an infectious disease physician.
Finally, as a control group, healthy donors (HD) matched for age and sex, with a negative nasopharyngeal swab for SARS-CoV-2 RNA detection, undetectable anti-SARS-CoV-2-specific IgG and without any symptoms, were enrolled.
2.2. Data and Sample Collection
An ad hoc electronic database was created to collect demographic data, comorbidities, laboratory results, oxygen support and type of ventilation. Two time points were considered: hospital admission (baseline) and post-COVID-19 visit at three months after discharge (Tpost). All blood tests were performed in the hospital’s central laboratory following standard procedures.
2.3. Microfluidic Next Generation Enzyme-Linked Immunosorbent Assay (ELISA)
During routine clinical testing, CSF samples were collected in sterile tubes without anticoagulant while peripheral whole blood was collected in heparin-coated BD Vacutainer Blood Collection tubes (BD Biosciences, Franklin Lakes, NJ, USA). All samples were collected between 8 and 10 a.m. As previously described [
2,
47], in collected samples, the evaluation of NfL, GFAP, sCD163, MMP-9 and TIMP-1 levels was assessed using the Simple Plex
TM Ella Assay (ProteinSimple, San Jose, CA, USA) on Ella
TM microfluidic system (Bio-Techne, Minneapolis, MN, USA) according to the manufacturer’s instructions. Ella
TM was calibrated using the in-cartridge factory standard curve. The limits of detection of NfL, GFAP, sCD163, MMP-9, and TIMP-1 were 1.09 pg/mL, 14.4 pg/mL, 318 pg/mL, 10.5 pg/mL, and 0.34 pg/mL, respectively. The limits of detection were calculated by adding three standard deviations to the mean background signal determined from multiple runs.
2.4. Statistical Analysis
Analysis was performed with Prism 9 (GraphPad, Boston, MA, USA). A probability value < 0.05 was considered as statistically significant and NfL, GFAP and sCD163 were examined as continuous variables. Patient characteristics were compared using Student’s t-test or chi-square for continuous and categorical variables, respectively. Continuous variables were expressed as the median and interquartile range (IQR) with the assumption of a normal distribution. Categorical variables were expressed as counts and percentages. Groups were then compared using Student’s t-test or the Mann–Whitney U-test, as appropriate. The nonparametric Kruskal–Wallis test with Dunn’s post-test was used for comparing medians of groups and subgroups with HD as well as for comparing medians of IOT, NIV and VMK with AA subgroup. The nonparametric Wilcoxon test was used for the longitudinal evaluation comparing baseline and Tpost. Correlations were performed using Spearman rank correlation analysis.
4. Discussion
In this study, to investigate the influence of hypoxia on the levels of neuronal and astrocytic injury and myeloid activation biomarkers, we stratified the study population according to both COVID-19 severity and maximum oxygen supply/ventilation support required during hospitalization. A further stratification was performed based on the presence of NS at both time points. Finally, an investigation of neuronal (NfL) and astrocyte damage (GFAP), myeloid activation (sCD163) and BBB permeability (MMP-9 and TIMP-1) was performed on the CSF samples of hospitalized COVID-19 patients with severe NS on hospital admission.
Our key findings are that: (1) on hospital admission, plasma levels of NfL, GFAP and sCD163 were significantly higher in COVID-19 patients compared to HD, especially in those who developed a severe form of the disease requiring ventilation support during hospitalization; (2) plasma levels of NfL and GFAP were significantly higher in patients with NS at the acute stage of the disease, compared to those without NS but with equal COVID-19 severity; (3) in COVID-19 patients with severe NS on hospital admission, positive correlations between CSF levels of sCD163 and CSF levels of NfL, GFAP and MMP-9 were observed; (4) plasma levels of NfL, GFAP and sCD163 were significantly reduced three months after hospital discharge, although levels were still higher compared to HD, especially in those with a severe form in the acute stage of COVID-19; (5) three months after hospital discharge, NS were self-reported by patients, and among them, patients with a severe form in the acute stage of COVID-19 showed higher plasma levels of sCD163 compared to those without NS. Conversely, no differences in plasma levels of NfL and GFAP were observed.
Consistent with previously published data, our results confirmed that plasma levels of CNS damage biomarkers were higher in hospitalized patients with severe COVID-19 [
18,
20,
48,
49,
50,
51]. According to the existing literature, in the acute stage of COVID-19 neurological involvement is accompanied by CNS damage, although the direct infection of brain parenchyma by SARS-CoV-2 remains a debated issue [
2,
52]. In addition to possible direct viral invasion, several mechanisms likely contribute to CNS involvement in COVID-19, including the indirect effects of systemic inflammation due to immune activation or hypoxia as a hematogenous pathway [
3,
4,
5]. To date, there is increasing evidence that neuronal injury is mediated primarily via hypoxia [
2] and hyperinflammation [
5]. Overall, our data confirmed that hypoxemia and systemic inflammation could be possible causes of CNS injury.
As a marker of inflammation, sCD163 is elevated in a range of diseases [
34,
43,
53,
54] and it was largely investigated in COVID-19 pneumonia [
42,
43,
44,
55]. As previously reported, a potential use to assess the risk of disease progression has been proposed, because an increase in plasma levels of sCD163 was observed on hospital admission in COVID-19 patients, especially in those who developed ARDS, as well as its correlation with typical inflammatory markers of COVID-19 pneumonia [
43,
44]. To the best of our knowledge, there is no information on the assessment of sCD163 in the CSF of COVID-19 patients nor on its correlation with NS at both the acute stage and post-COVID. However, our data seem to confirm the potential role of sCD163 as a CNS impairment biomarker. sCD163 is known to be upregulated during the pro-inflammatory response, and the release of MMPs is known to contribute to this orchestration [
56]. Thus, despite the small sample size, we observed a positive correlation between the CSF levels of sCD163 and MMP-9, underlining that increased CSF levels of sCD163 and MMP-9 might also contribute to the infiltration of monocytes to the CSF in COVID-19 patients. In addition, the positive correlations between CSF levels of sCD163 and CSF levels of both NfL and GFAP underline the ongoing inflammation, as postulated in other neurological conditions [
34]. Indeed, sCD163 levels in the CSF have been shown to be elevated in patients with multiple sclerosis as well as in combination antiretroviral therapy-treated people living with HIV with a mild neurocognitive disorder. Plasma sCD163 levels were elevated compared to those who are cognitively normal or who have asymptomatic neurocognitive impairment [
57]. Also, in neurodegenerative disorders, such as Parkinson’s disease, sCD163 is a potential cognition-related biomarker underlighting a role of monocytes in both peripheral and brain immune responses [
35].
Finally, positive correlations between the CSF and plasma levels of NfL and GFAP in COVID-19 patients with severe NS on hospital admission, validate their plasma assessment as a less-invasive biomarker for diagnosis, prognosis and monitoring of CNS damage associated with COVID-19. Moreover, activated astrocytes upregulate GFAP expression, produce fine processes, and exhibit hypertrophic morphology [
58]. In addition to the role of active astrocytes in the formation of the physical BBB, aberrant ECM proteins at the lesion site also inhibit reparative precursor cell migration during recovery [
59,
60]. MMP-9 plays a key role in ECM remodeling during brain injury due to its capacity to proteolytically degrade ECM [
61]. Our data underline the detrimental effects of MMP-9, including BBB disruption and inflammation, in the acute phase of COVID-19 associated with astrocyte damage.
To clarify the role of NfL, GFAP and sCD163 as a sign for CNS damage presenting with NS, we stratified hospitalized COVID-19 patients according to the severe presence of NS during the acute stage of COVID-19 and self-referred NS three months after hospital discharge. During the acute stage of the disease, patients with severe NS (requiring puncture lumbar for diagnostic purposes) showed higher plasma levels of NfL, GFAP and sCD163 compared to patients without severe NS. However, three months after hospital discharge, no differences in plasma levels of both NfL and GFAP between patients with and without self-referred NS were found. However, among severe COVID-19 patients, the high plasma levels of NfL and GFAP in patients without severe NS suggest the presence of subclinical central nervous system involvement. Conversely, among patients with severe COVID-19 at the acute stage, high plasma levels of sCD163 seem to be associated with self-referred NS. The elevated levels of sCD163 could be a mechanism of tissue homeostasis and repair and thus sCD163 could be a marker of immune modulatory functions regarding not only degeneration and anti-inflammation but also tissue repair [
62,
63,
64].
Finally, the longitudinal evaluation showed that despite the significant reduction in plasma levels of CNS injury biomarkers, three months after hospital discharge, COVID-19 patients still have persistently higher levels compared to HD. These data are in line with previous reports [
2,
65] although Kanberg et al. [
50] reported a complete normalization of CNS injury biomarkers after six months post-infection.
Our single-center study has limitations such as the small sample size of patients with severe NS at the acute stage of COVID-19 and the evaluation of self-referred NS three months after hospital discharge, which was not confirmed by standardized cognitive tests but recorded by an infectious disease physician. Finally, only half of the enrolled patients were evaluated at the post-COVID clinic and among them just 5 out 15 patients had severe NS at the acute stage of COVID-19.
Overall, plasma biomarkers of brain injury, NFL and GFAP, as well as monocyte/macrophage activation markers have been found to be increased in a severity-dependent manner in hospitalized COVID-19 patients. High plasma and CSF levels of NfL and GFAP in COVID-19 could be due to a proinflammatory systemic and brain response that involves microglial activation and subsequent neuronal damage. Our data further highlight the association between myeloid activation and CNS perturbations.
The recognition and diagnosis of these neurologic complications at both acute-stage and post-COVID are challenging, particularly in the context of overstrained medical systems, where an under-recognition or delays in diagnosis may contribute to poor outcomes [
12]. However, further studies are required to clarify the nature of CNS injury and evaluate the usefulness of these biomarkers in COVID-19 patients.