key: cord-0998067-d14tpdny authors: Ogata, Alana F; Maley, Adam M; Wu, Connie; Gilboa, Tal; Norman, Maia; Lazarovits, Roey; Mao, Chih-Ping; Newton, Gail; Chang, Matthew; Nguyen, Katrina; Kamkaew, Maliwan; Zhu, Quan; Gibson, Travis E; Ryan, Edward T; Charles, Richelle C; Marasco, Wayne A; Walt, David R title: Ultra-sensitive Serial Profiling of SARS-CoV-2 Antigens and Antibodies in Plasma to Understand Disease Progression in COVID-19 Patients with Severe Disease date: 2020-09-08 journal: Clin Chem DOI: 10.1093/clinchem/hvaa213 sha: 9176c5b2fd8a31ca42ec4d9e0d2860b4cb9ecb4b doc_id: 998067 cord_uid: d14tpdny BACKGROUND: Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has infected over 21 million people worldwide since August 16, 2020. Compared to PCR and serology tests, SARS-CoV-2 antigen assays are underdeveloped, despite their potential to identify active infection and monitor disease progression. METHODS: We used Single Molecule Array (Simoa) assays to quantitatively detect SARS-CoV-2 spike, S1 subunit, and nucleocapsid antigens in the plasma of coronavirus disease (COVID-19) patients. We studied plasma from 64 COVID-19 positive patients, 17 COVID-19 negative patients, and 34 pre-pandemic patients. Combined with Simoa anti-SARS-CoV-2 serological assays, we quantified changes in 31 SARS-CoV-2 biomarkers in 272 longitudinal plasma samples obtained for 39 COVID-19 patients. Data were analyzed by hierarchical clustering and were compared to longitudinal RT-PCR test results and clinical outcomes. RESULTS: SARS-CoV-2 S1 and N antigens were detectable in 41 out of 64 COVID-19 positive patients. In these patients, full antigen clearance in plasma was observed a mean ± 95%CI of 5 ± 1 days after seroconversion and nasopharyngeal RT-PCR tests reported positive results for 15 ± 5 days after viral antigen clearance. Correlation between patients with high concentrations of S1 antigen and ICU admission (77%) and time to intubation (within one day) was statistically significant. CONCLUSIONS: The reported SARS-CoV-2 Simoa antigen assay is the first to detect viral antigens in the plasma of COVID-19 positive patients to date. These data show that SARS-CoV-2 viral antigens in the blood are associated with disease progression, such as respiratory failure, in COVID-19 cases with severe disease. The current pandemic of coronavirus disease , caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has resulted in over 21,000,000 confirmed cases globally and over 167,000 deaths in the United States alone, as of August 16, 2020 To address the need for a quantitative antigen assay, we developed ultra-sensitive Single Molecule Array (Simoa) SARS-CoV-2 antigen assays for S1, S1-S2 extracellular domain (spike), and nucleocapsid (N). The ultra-sensitivity of Simoa enables detection of SARS-CoV-2 antigens in the plasma of COVID-19 positive patients. Additionally, Simoa provides a dynamic range that allows quantification of antigens over a concentration range of four orders of magnitude. This precise quantification is advantageous for capturing the wide range of antigen concentrations in COVID-19 patient plasma throughout the course of hospitalization. We combined our Simoa SARS-CoV-2 antigen assays with previously developed Simoa serological assays to monitor SARS-CoV-2 antigens and anti-SARS-CoV-2 immunoglobulins in longitudinal plasma samples of COVID-19 patients. These measurements provide direct evidence of the inverse correlation between anti-SARS-CoV-2 antibody production and viral antigen clearance from plasma, which provides a unique view of viral infection and immune response from the beginning of hospitalization through recovery or death. All plasma was collected in purple top K2-EDTA tubes and centrifuged at 2,000 × g at 4°C for 10 min prior to analysis. COVID-19 positive and negative samples were obtained from adult patients presenting to Brigham and Women's Hospital or Massachusetts General Hospital. We received 17 samples from patients who tested negative for SARS-CoV-2 using NP RT-PCR. We received 64 samples from patients who tested positive for SARS-CoV-2 using NP RT-PCR. We Preparation of viral antigen assay Simoa reagents is in the online Supplemental Materials. Simoa assays were performed on an HD-X Analyzer (Quanterix) in an automated three-step assay format according to the manufacturer's instructions and as previously described. (8) Plasma samples were diluted 8-fold in Homebrew Detector/Sample Diluent (Quanterix) with Halt Protease Inhibitor Cocktail (ThermoFischer Scientific) and EDTA. Detector antibodies were diluted in Homebrew Detector/Sample Diluent to 0.3 µg/mL, and streptavidin-β-galactosidase (SβG) concentrate (Quanterix) was diluted to 150 pmol/L in SβG Diluent (Quanterix). Antibodyconjugated capture beads were diluted in Bead Diluent, with a total of 500,000 beads per reaction (125,000 S1 beads, 125,000 S2 beads, and 250,000 647 nm dye-encoded helper beads for the S1/S2 multiplex assay, and 125,000 nucleocapsid beads and 375,000 647 nm dyeencoded helper beads for the nucleocapsid assay). All reagents were diluted in plastic bottles that were loaded into the HD-X Analyzer, and all assay steps were performed in an automated manner on the instrument. In each assay, capture beads were incubated with the sample for 15 minutes, detector antibody for 5 minutes, and SβG for 5 minutes, with washing steps in between. The beads were then resuspended in 25 µL resorufin-β-galactopyranoside and loaded into the microwell array for imaging. Average enzyme per bead (AEB) and sample concentration values were calculated by the HD-X Analyzer software. All samples were measured in duplicates. Immunoglobulin assays were performed using a similar procedure (online Supplemental Materials). Seroconversion classification was determined based upon the early stage classification model (9) trained using an independent panel of 142 samples positive by RT-PCR SARS-Cov-2 and 200 negative pre-pandemic controls. The markers for this model were chosen using a five-fold crossvalidation step as previously reported. (9) Two cross validations were run: 1) Early-stage cases and pre-pandemic controls; 2) late-stage cases and all controls. Each training set initially considered all 12 markers (IgG, IgM, and IgA each against S1, Spike, N, and RBD). This crossvalidation yielded four markers (IgA S1, IgA Nucleocapsid, IgG Nucleocapsid, and IgG Spike) and exhibited the best performance in the training set. The threshold for a positive test result for the unknown samples was determined based on the cutoff that yielded 100% specificity in the training set. The cluster map analysis was performed on standardized biomarker data. An example (online For SARS-CoV-2 antigen Simoa assays, three types of dye-encoded paramagnetic beads were functionalized with antibodies against each viral antigen and incubated with plasma samples for multiplexed Simoa measurements (Fig. 1a , Methods) as described previously. (8) The SARS-CoV-2 Simoa antigen assays detect S1, spike, and N antigens with limits of detection (LOD) of 5 pg/mL (0.07 pmol/L), 70 pg/mL (0.39 pmol/L), and 0.02 pg/mL (0.4 fmol/L), respectively (online Supplemental Table S2 ). The range in LODs for each viral antigen assay is primarily due to differences in affinities of the capture and detection antibodies, as antibody dissociation rate constants are an important factor for Simoa assay sensitivity. Table S3 ). We measured anti-SARS-CoV-2 immunoglobulins (total IgA, IgM, and IgG) using our recently established SARS-CoV-2 serological assays (8) to correlate viral antigens with immunoglobulin levels (Fig. 1a) . In addition, we developed SARS-CoV-2 serological Simoa assays for IgG1, IgG2, IgG3, and IgG4 detection (online Supplemental Materials). The combined measurements of three SARS-CoV-2 antigens and seven anti-SARS-CoV-2 immunoglobulin isotypes against four SARS-CoV-2 antigens enables quantification of 31 biomarkers from 70 µL of a plasma sample. To probe for the presence of viral antigens in plasma, we tested samples from COVID-19 positive patients using our SARS-CoV-2 Simoa antigen assays for S1, spike, and N. These patients were determined to be COVID-19 positive by NP RT-PCR and all plasma samples were obtained within the first ten days of the initial NP RT-PCR test. Corresponding immunoglobulin levels are presented in online Supplemental Figures S4 and S5 . S1 and N were detected in 41 of 64 COVID-19 positive patients ( Fig. 1b and 1c) , who we identify as "viral antigen positive." Despite the presence of S1 and N in some samples, spike was only detectable in 5 of 64 COVID-19 positive patients (online Supplemental Figure S6 ). Spike may be undetectable in some samples since the LOD is one order of magnitude higher than the LOD of the S1 assay. Additionally, in the Simoa assay for spike, the formation of a full immunocomplex depends on spike binding to the S2 subunit capture beads and the S1 subunit detection antibody. Therefore, we hypothesize that free spike antigen in plasma is likely proteolytically cleaved, releasing the S1 subunit, and the remaining spike protein fragment is undetectable by our assay. Cross-reactivity of the SARS-CoV-2 antigen assays was assessed using plasma samples from three control patient cohorts: (1) samples from individuals who tested negative for COVID-19 by NP RT-PCR, (2) pre-pandemic samples from healthy individuals, and (3) (Fig. 1b, c) . However, based on our preliminary results, a multiplexed approach for the detection of S1 and N can improve analytical specificity of these assays. A more detailed discussion is provided in the online Supplemental Materials. We attribute detection of viral antigens in COVID-19 negative patients to either (1) assay cross-reactivity with other coronaviruses or (2) Figures S10, S12-14, S19-20, S23, S26, S31, S34, S36) . As shown in Fig. 2 , among Patients 21, 22, and 26, once a patient has seroconverted and their plasma immunoglobulin levels reach a steady state, there are typically no detectable concentrations of viral antigen. All patients seroconverted a mean ± (95% CI) of 7 ± 1 days after the first NP RT-PCR positive test, in agreement with previous serological studies. (13) (14) (15) (16) We defined viral antigen clearance as the first day that both S1 and N were undetectable in patient's plasma. For viral antigen positive patients, full antigen clearance in plasma was observed 5 ± 1 days after seroconversion. were already seroconverted at their first NP RT-PCR test (Patient 39 in Fig. 2 and online Supplemental Figures S11, S15-17, S18, S24, S27-28, S30-33, S35, S7-44) . We propose three possibilities for the lack of detectable antigen in some patient plasma: (1) patients are presenting to the hospital after seroconversion, and therefore most viral antigens have been cleared from the plasma, (2) we hypothesize that COVID-19 cases with very severe disease will have viral antigen leakage into the blood, and therefore patients that do not progress to more severe forms of the disease will not have viral antigen leakage into the blood, or (3) viral antigens are present but are bound to an immunoglobulin, which blocks a binding epitope of the capture or detection antibody, resulting in an antigen-immunoglobulin complex that is undetectable by our Simoa assays. However, even with no detectable concentrations of viral antigen, 12 of 16 patients were admitted or transferred to the ICU during hospitalization and intubated, indicating that these patients were severe cases. Therefore, we propose a third possibility where patients have seroconverted and despite reaching viral clearance, suffer from severe respiratory damage that inhibits recovery. patients. The cluster map shows three dominant branches (Fig. 3) IgG1 and IgG3 are subclasses of IgG that predominantly respond to viral antigens and mediate neutralization. (17) (18) (19) (20) (21) Based on these measurements, the IgG response that was mounted by the immune system was primarily mediated by IgG1 and IgG3 for SARS-CoV-2, in agreement with previous serological studies. (4, 5) The neutralization effects of IgG1 and IgG3 on viral antigens should be explored in future studies that include correlative neutralization titer assays. We next explored how antigen clearance compared with longitudinal RT-PCR tests and clinical outcomes. The measured SARS-CoV-2 antigen and immunoglobulin levels in plasma, NP RT-PCR tests, and select clinical outcomes for each of these patients are displayed in Fig. 4 and patient features such as age and gender are summarized in online Table S4 . In this patient cohort, NP RT-PCR tests reported positive results for a mean of 15 ± 5 days after viral antigen clearance and 18 ± 4 days after seroconversion. Furthermore, several patients never received a negative RT-PCR result before being discharged from the hospital. These observations are in agreement with recent virological studies that confirmed persistence of viral RNA and showed how seroconversion was not immediately followed by a decrease in NP viral RNA in hospitalized patients. (22) Because RNA shedding can occur for several weeks after infection and recovery, measurements of viral antigens and immunoglobulins may provide more timely indicators of infectivity compared to viral RNA, but additional studies such as viral load measurements are necessary to corroborate these findings. To understand the correlation between SARS-CoV-2 antigen concentrations and disease severity, clinical features for the 64 COVID-19 positive patients (Fig. 5 , patient IDs 1-64) were compared with viral antigen concentrations in plasma. It is important to note that all samples were obtained from patients admitted to the hospital, resulting in a cohort of cases primarily with severe disease. S1 shows higher correlation with clinical severity than N in plasma (online Supplemental Table S5 -8, Figure S45 ). Confounding factors such as age and sex were also assessed and showed no correlation with clinical outcomes (online Supplemental Table S9 , S10 and Figure S46) . Therefore, patients were grouped into three categories of S1 concentrations: (1) 23 patients with undetectable S1 concentrations (below the limit of detection), (2) 23 patients with low concentrations of S1 (6-50 pg/mL, 0.08-0.65 pmol/L), and (3) 18 patients with high concentrations of S1 (>50 pg/mL, > 0.65 pmol/L). There is a significant difference in rates of ICU admission upon presentation for the three patient groups based on S1 concentrations in plasma (P-value: 0.0107). Patients with zero, low, and high concentrations of S1 were admitted to the ICU upon presentation to the hospital at rates of 30% (7 of 23 patients), 52% (12 of 23 patients), and 77% (14 of 18 patients), respectively (Fig. 5a) . Among all COVID-19 positive patients, over 60% were intubated during hospitalization with no statistically significant difference in intubation rates among groups (Fig. 5b) . The difference in mean times to intubation between patients with high concentrations of S1 and patients with no detectable S1 is significant (P-value: 0.0050), where all patients with high concentrations of S1 were intubated within one day of hospitalization (Fig. 5c) . These results suggest that high S1 concentrations in plasma upon presentation to the hospital correlate with severe cases of COVID-19 that can result in respiratory failure and require immediate intubation. Among all patients, a broad range of intubation periods, up to 32 days, was observed (Fig. 5d) . Patients grouped by S1 concentrations, age, and sex showed no statistically significant difference in death rate (online Supplemental Table S9 , S10 and Figure S46 ). Although six patients showed detectable concentrations of spike protein, there was no correlation between spike concentrations and ICU admission, intubation rate, or death rate. There are potentially more features in these data that will lead to further correlations, but a large cohort of patients that include asymptomatic and mild cases will need to be tested to elucidate other patterns. Plasma is a readily available biofluid from hospitalized patients but is more difficult to obtain in non-clinical settings. In comparison, saliva is a non-invasive biofluid that is easier to use for wide-scale testing and may be more interesting for monitoring viral antigen concentrations to understand active viral infection from COVID-19, which is a respiratory disease. We tested 17 saliva samples from patients presenting to the Emergency Department at Brigham and Women's Hospital who were tested by NP RT-PCR for COVID-19 (online Supplemental Figure S47, S48, S49) . We found that S1 and N was detectable in 7 of 11 COVID-19 patients when compared to healthy saliva controls. Similar to our observation of antigen and immunoglobulins in plasma, there were 7 of 11 patients with low S1 concentrations and high levels of IgA-S1. One patient showed a background concentration of IgA-S1 and notably high concentration of S1 (135 pg/mL). However, a deep understanding of the correlation between viral antigens and immunoglobulins in saliva will require a larger sample cohort. Nonetheless, these initial results indicate the presence of SARS-CoV-2 antigens and anti-SARS-CoV-2 immunoglobulins in saliva and highlight the potential for adapting our assays to a diagnostic test for COVID-19. Future studies on saliva will include longitudinal sample analysis for mild and severe cases of COVID-19 patients and will explore the potential for developing a saliva-based COVID-19 antigen screening tool. Using SARS-CoV-2 Simoa assays, we have demonstrated quantitative detection of SARS-CoV-2 antigens and anti-SARS-CoV-2 immunoglobulins in plasma of COVID-19 patients. While detection of N in NP swabs has been cited, (6) we present the first report of SARS-CoV-2 spike, S1, and N detection in plasma. The presence of S1 and N in plasma suggest that fragments of virus are entering the bloodstream, potentially due to tissue damage. Although spike is undetectable in most COVID-19 patients, possibly due to proteolytic cleavage, six patients showed high concentrations of spike in plasma. No evidence has been reported yet for full viral particles in blood, though we cannot rule out this possibility.(23) Nonetheless, severe COVID-19 cases with acute respiratory distress syndrome can result in damage to endothelial cells and vascular leakage (24) (25) (26) (27) and we propose that this damage can lead to discharge of viral antigens into the blood. Patients with lung damage can suffer from respiratory failure and require intubation or mechanical ventilation. We also found significant correlation between high S1 concentrations in plasma and time between hospital admission and intubation. Although we hypothesize that asymptomatic or mild cases will likely not show viral antigen concentrations in plasma, future studies that include mild cases will be used to probe SARS-CoV-2 antigen and antibody concentrations over time in comparison to the severe cases presented here. Notes: 1. World Health Organization. Coronavirus disease (COVID-19) Situation Report 178 Interpreting Diagnostic Tests for SARS-CoV-2 Systemslevel immunomonitoring from acute to recovery phase of severe COVID-19 Detection of SARS-CoV-2-Specific Humoral and Cellular Immunity in COVID-19 Convalescent Individuals A serological assay to detect SARS-CoV-2 seroconversion in humans Clinically practiced and commercially viable nanobio engineered analytical methods for COVID-19 diagnosis A fullyautomated, six-plex single molecule immunoassay for measuring cytokines in blood Ultra-Sensitive High-Resolution Profiling of Anti-SARS-CoV-2 Antibodies for Detecting Early Seroconversion in COVID-19 Patients The SciPy Community. Scipy-cluster-hierarchy-linkage Using antigen-antibody binding kinetic parameters to understand single-molecule array immunoassay performance False Negative Tests for SARS-CoV-2 Infection -Challenges and Implications Antibody responses to SARS-CoV-2 in patients with COVID-19 A serological assay to detect SARS-CoV-2 seroconversion in humans A systematic review of antibody mediated immunity to coronaviruses: antibody kinetics, 18 correlates of protection, and association of antibody responses with severity of disease Convergent Antibody Responses to SARS-CoV-2 Infection in Convalescent Individuals IgG subclasses and allotypes: From structure to effector functions Effector functions of camelid heavy-chain antibodies in immunity to West Nile virus Kinetics of dengue virus-specific serum immunoglobulin classes and subclasses correlate with clinical outcome of infection Immunoglobulin G3 from Polyclonal Human Immunodeficiency Virus (HIV) Immune Globulin Is More Potent than Other Subclasses in Neutralizing HIV Type 1 Immunoglobulin G subclasses in antibody responses to St. Louis encephalitis virus infections Virological assessment of hospitalized patients with COVID-2019 Detection of SARS-CoV-2 in Different Types of Clinical Specimens Complement associated microvascular injury and thrombosis in the pathogenesis of severe COVID-19 infection: a report of five cases Outbreak of SARS-CoV2: Pathogenesis of infection and cardiovascular involvement Dysregulated Type I Interferon and Inflammatory Monocyte-Macrophage Responses Cause Lethal Pneumonia in SARS-CoV-Infected Mice COVID-19: the vasculature unleashed Serial data for four individual COVID-19 positive patients from admission to clinical recovery or death. Simoa antigen and serological results for serial plasma samples. (1) the total IgA, IgM, and IgG levels against four viral antigens: nucleocapsid (N), receptor binding domain (RBD), S1, and spike, for a total of 12 interactions S1 antigen concentrations with IgG1-4 against S1. Data points represent the mean concentration or AEB from two replicate measurements and error bars represent the standard error of the mean from two replicate measurements The authors acknowledge Dr. Sanjat Kanjilal for collecting saliva samples. A version of this paper was previously posted as a preprint on medRxiv as https://www.medrxiv.org/content/10.1101/2020.07.20.20156372v1.