key: cord-262640-4vr4cm1s authors: Nguyen, N. N.; Mutnal, M. B.; Gomez, R. R.; Pham, H. N.; Nguyen, L. T.; Koss, W.; Rao, A.; Arroliga, A. C.; Wang, L.; Wang, D.; Hua, Y.; Powell, P. R.; Chen, L.; McCormack, C.; Linz, W. J.; Mohammad, A. A. title: Correlation of ELISA based with random access serologic immunoassays for identifying adaptive immune response to SARS-CoV-2 date: 2020-07-08 journal: nan DOI: 10.1101/2020.07.06.20145938 sha: doc_id: 262640 cord_uid: 4vr4cm1s Public health emergency of SARS-CoV-2 has facilitated diagnostic testing as a related medical countermeasure against COVID-19 outbreak. Numerous serologic antibody tests have become available through an expedited federal emergency use only process. This paper highlights the analytical characteristic of an ELISA based assay by AnshLabs and three random access immunoassay (RAIA) by DiaSorin, Roche, and Abbott that have been approved for emergency use authorization (EUA), at a tertiary academic center in a low disease-prevalence area. The AnshLabs gave higher estimates of sero-prevalence, over the three RAIA methods. For positive results, AnshLabs had 93.3% and 100% concordance with DiaSorin or Abbott and Roche respectively. For negative results, AnshLabs had 69.7% and 73.0% concordance with DiaSorin and Roche or Abbott respectively. All discrepant samples that were positive by AnshLabs and negative by RAIA tested positive by all-in-one step SARS-CoV-2 Total (COV2T) assay performed on the automated Siemens Advia Centaur XPT analyzer. None of these methods, however, are useful in early diagnosis of SARSCoV- 2. The SARS-CoV-2 virus outbreak that began in late 2019 in Wuhan, has a mortality rate of approximately 46 6.1% worldwide [1] [2] [3] . Diagnostic testing is necessary for identifying and isolating infected individuals to 47 limit spread of disease. Molecular testing such as reverse-transcriptase polymerase chain reaction (rtPCR) 48 detects active infection; and serology testing helps identify those who were previously infected (including 49 asymptomatic infections) and have recovered [4, 5] . Nucleic acid detection using rtPCR has become the 50 confirmation test, due to its 99% specificity and 60-90% sensitivity within 7 days of exposure [6] but is 51 faced with numerous supply challenges [7] . The United States Food and Drug Administration (FDA) 52 issued an Emergency Use Authorization (EUA) approval for antibody testing as complementary to rtPCR, 53 leading to an explosion of new antibody methods, including rapid diagnostic test (RDT), enzyme-linked 54 immunosorbent assay (ELISA), virus neutralization assay (VNA), and chemiluminescent immunoassay 55 One hundred and fifty-two serum samples were from patients who tested negative by rtPCR, 134 of these 80 were collected on same day as rtPCR testing. For the remaining 18 samples, the interval between rtPCR 81 and sample collection ranged from 1-48 days. To avoid degradation, the specimens were tested by four 82 methodologies within 12-20 h of each other. Only samples having sufficient serum volume and rtPCR test 83 results were included in the evaluation project. 84 85 Table 1 summarizes the characteristics of the four serologic assays we investigated. 86 peroxidase. After a second incubation and washing step, the wells are incubated with the substrate 97 tetramethylbenzidine (TMB) chromogen solution to induce color change. An acidic stopping solution is 98 All rights reserved. No reuse allowed without permission. was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which this version posted July 8, 2020 . . https://doi.org/10.1101 added and the degree of enzymatic turnover of the substrate is determined by wavelength absorbance 99 measurement, with 450 nm as the primary filter and 630 nm as the reference filter. The intensity of color 100 change corresponds to arbitrary units of antibody-antigen complex concentration present in the specimen. 101 The analyzer calculates antibody concentration in arbitrary concentration units (AU/mL). Samples with 102 AU/mL of >12, 10-12, and <10 are considered positive, indeterminate and negative for IgG respectively. 103 It is the only test that uses a three-point calibration curve. The sensitivity and specificity are 95.0% and 104 98.3% respectively [11] . 105 The Abbott SARS-CoV-2 IgG assay was run on the Abbott Architect i2000SR analyzer that measures 106 IgG antibodies to the nucleocapsid protein. The automated, two-step immunoassay uses 107 chemiluminescent microparticle immunoassay (CMIA) technology for qualitative detection of IgG 108 antibodies in human serum. The sample, SARS-CoV-2 antigen-coated paramagnetic microparticles, and 109 diluent are combined and incubated. The antibodies bind to the antigen-coated microparticles. The 110 mixture is washed and anti-human IgG acridinium-labeled conjugate is added. Following incubation, the 111 pre-trigger is added. The resulting chemiluminescent reaction is measured as a relative light unit (RLU). 112 The presence or absence of IgG antibodies is determined by dividing the sample RLU by the stored 113 calibrator RLU to find the IgG assay index (S/C), with a positive cutoff of ≥1.4. The sensitivity and 114 specificity are 100% and 99.63% respectively at ≥ 14 days post onset of symptoms [12] . 115 The LIAISON SARS-Cov-2 S1/S2 IgG is a chemiluminescent immunoassay (CLIA) for detection of anti-116 S1 and anti-S2 spike glycoprotein specific to SARS-CoV-2 in human serum or plasma on the DiaSorin 117 XL analyzer (Stillwater, MN). Specimen, calibrator, control, coated magnetic particles and diluent are 118 incubated in reaction cuvettes. The antibodies bind to the solid phase through the recombinant S1 and S2 119 antigens. A second incubation links recombinant S1 and S2 antigens to an isoluminol-antibody conjugate. 120 The starter reagents are then added, and a flash chemiluminescence reaction induced. The light signal, and 121 hence the amount of isoluminol-antibody conjugate, is measured by a photomultiplier and result 122 converted to arbitrary concentration, AU/mL. Samples with AU/mL of ≥15 are considered positive for 123 All rights reserved. No reuse allowed without permission. was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which this version posted July 8, 2020. . https://doi.org/10.1101/2020.07.06.20145938 doi: medRxiv preprint IgG antibodies. The sensitivity and specificity are 90-97% and 98% respectively ≥ 14 days post onset of 124 symptoms [13] . 125 The Elecsys Anti-SARS-CoV-2 assay is performed on the Roche cobas e601 analyzer for total antibodies 126 specific for IgG, IgM and IgA which target nucleocapsid protein, in human serum or plasma. In order to rule out non-specific binding, samples that tested positive by ELISA assay were diluted using 144 sample diluent provided in the AnshLabs assay kit. We made and reran samples for a 1:2, 1:3 and 1:4 145 dilution, and calculated percent recovery. 146 All rights reserved. No reuse allowed without permission. was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which this version posted July 8, 2020. The specificities of the validated in-house AnshLabs SARS-CoV-2-IgG and IgM are listed in Table 2 . 156 The cross reactivity to anti-influenza B IgG (5 samples), anti-respiratory syncytial virus IgG (5 samples were positive for these analytes. No cross-reactivity was noted for either SARS-CoV-2-IgG or IgM. The 161 clinical sensitivity and specificity using rtPCR results as the gold standard were found to be 86.7and 162 91.2% respectively. All samples used for the sensitivity and specificity evaluation were collected from 163 symptomatic patients, either hospitalized inpatients or treated in Emergency Department. The interval 164 between rtPCR confirmation and serology testing ranged from 2-12 days. The specificity toward common 165 cold coronavirus is shown in Table 3 . Three of 100 prepandemic samples tested positive for IgG by 166 ELISA and none tested positive by RAIA methods, thereby giving a calculated specificity of 97% and 167 100% for ELISA and RAIA respectively. 168 All rights reserved. No reuse allowed without permission. was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which this version posted July 8, 2020. . https://doi.org/10.1101/2020.07.06.20145938 doi: medRxiv preprint Table 4 shows the concordance between ELISA and RAIA results for samples that were confirmed 174 positive for SARS-CoV-2 by rtPCR. These samples were collected from symptomatic patients > 13 days 175 post rtPCR confirmation. ELISA assay correlated best with Total Antibody assay on Roche Elecsys e601 176 analyzer. This could possibly be attributed to the measurement of IgG antibodies directed towards 177 multiple antigenic proteins (nucleocapsid & spike) by ELISA or measurement of total antibodies (IgG, 178 IgM, and IgA) on Roche Elecsys e601 analyzer. 179 Table 4 . Concordance of 15 rtPCR positive samples between a) ELISA and RAIA systems and b) among 180 three RAIA platforms 181 AnshLabs IgG vs Architect i2000 93.3% AnshLabs IgG vs Elecsys e601 100% All rights reserved. No reuse allowed without permission. was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which this version posted July 8, 2020. The concordance of ELISA and RAIA results with rtPCR is shown in Table 6 . All patient tested positive 193 by rtPCR also tested positive by ELISA and Elecsys e601 total antibody. Architect i2000 IgG and Liaison XL were unable to detect antibodies in one sample. All RAIA methodologies showed 195 high correlation with nucleic acid test for patient samples that tested negative by rtPCR, with 196 All rights reserved. No reuse allowed without permission. was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which this version posted July 8, 2020. The non-specific binding dilution data of the AnshLabs assay showed five samples with various 202 concentration levels of IgG were serial diluted to 1:2, 1:4: 1:8 and 1:16. All samples gave a consistent 203 dilution pattern and expected 90-100% recovery of neat sample in AU/mL units (Fig 1) . 204 205 All RAIA methods correlated well with ELISA and rtPCR for samples collected >13 days post rtPCR 207 confirmation. There were no significant differences among the methods which tested for IgG targeted to 208 one or both nucleocapsid and spike proteins, or tested for total antibodies. 209 ELISA detected higher sero-prevalence in rtPCR negative samples than the RAIA methods. This may be 210 due to i) higher analytical sensitivity or a lower cutoff by ELISA, which triggered more positive results; 211 ii) cross reactivity to other coronavirus; iii) non-specific binding of other antibodies, for example 212 autoimmune antibodies or deposition of detection antibody on the microtiter well which led to increased 213 absorbance causing false positives 214 All rights reserved. No reuse allowed without permission. was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which this version posted July 8, 2020 . . https://doi.org/10.1101 ELISA assays are generally known for low detection limits in sub ng/mL to low pg/mL because of their 215 increased incubation time thereby allowing antigen-antibody to reach reaction equilibrium and extra 216 washing steps [15, 16] . The Dynex DSX analyzer used for ELISA assay provided optimization flexibility 217 and automation, which is not available on RAIA due to throughput constraint. Cross-reactivity to other 218 coronavirus was evaluated by testing 100 prepandemic samples and found to be 3% and 0% for ELISA 219 and RAIA respectively. The differences in cross-reactivity may account for one or two false positive 220 results, but not for all 34 and 15 positives picked up by ELISA. Non-specific deposition of other 221 antibodies in patient samples or detection antibody was ruled out by dilution studies for ELISA. Recovery 222 of 90-110% ruled out non-specific binding as a possible cause for false positives (Fig 1) . The difference 223 in results for positive and negative samples by RAIA methods may also be due to a higher threshold for 224 positivity. The rtPCR assay is used as the gold standard in maximizing analytical sensitivity and 225 specificity during method development which is the most accurate in the early days of the infection when 226 antibody development is low and results in the reported sensitivity of 10-60% on samples collected <14 227 days post rtPCR confirmation [17] [18] [19] . 228 We believe that higher rate of positivity observed for ELISA i.e. 34 versus 1 by Architect, 7 by Liaison was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which this version posted July 8, 2020. . https://doi.org/10. 1101 Our quality assurance project has some notable limitations. At this stage of the disease, true clinical 238 sensitivity and specificity for different methodologies is difficult to determine because of our limited 239 understanding of the disease process and kinetics. Secondly, our assumption that ELISA has better limits 240 of detection is based on circumstantial evidence, as certified standards quantifying limits of detection on 241 different platforms are not available. Third, the cutoffs provided by manufacturers were relied on which 242 may not have undergone extensive validation. Establishing laboratory specific cut-off is akin to 243 establishing reference ranges, which is highly dependent on prevalence of disease in local population. 244 245 All of the assays we investigated would work well for epidemiological sero-prevalence studies. Among 247 rtPCR negative patients, ELISA gave higher estimates of sero-prevalence in our dataset and would 248 probably do so in population-based epidemiological surveys using serological testing. RAIA methods 249 could however offer other advantages over ELISA which includes i) faster turnaround time; ii) random 250 access to allow immediate testing; iii) longer calibration stability, obviating the need to perform daily 251 calibration as required by ELISA; iv) the ability to perform other immunoassay testing concurrently; and 252 v) higher test throughput and walk away capabilities. However in conclusion, no serological method 253 tested has sensitivity and specificity greater than or equal to 99% for one to 5 days post exposure, limiting 254 their use in early diagnosis. 255 The authors would like to thank Ms. Briget Da Graca for editorial comments and manuscript revision. We 257 also acknowledge Ms. Laura Gonazales and her team from Health Texas Provider Network (Dallas, TX) 258 for correlation of AnshLabs results using Siemens Centaur total antibody assay. 259 260 All rights reserved. No reuse allowed without permission. was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which this version posted July 8, 2020. . https://doi.org/10.1101/2020.07.06.20145938 doi: medRxiv preprint non-specific binding in AnshLabs ELISA assay 320 All rights reserved. No reuse allowed without permission. was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which this version posted July 8, 2020. . https://doi.org/10.1101/2020.07.06.20145938 doi: medRxiv preprint Novel Coronavirus in the United States 264 2. World Health Organization. WHO Coronavirus Disease (COVID-2019) Dashboard The New England journal of medicine specific antibody responses in COVID-19 patients Antibody responses to SARS-CoV-2 in 273 patients of novel coronavirus disease 2019 Infectious Diseases Society of America Recent advances and 276 perspectives of nucleic acid detection for coronavirus European Centre for Disease Prevention and Control. Novel coronavirus disease 2019 (COVID-279 19) pandemic: increased transmission in the EU/EEA and the UK -sixth update -12 Evaluation 282 of nine commercial SARS-CoV-2 immunoassays Neutralizing antibody responses to SARS-285 Korean Center for Disease Control and Prevention. Findings from investigation and analysis of 288 re-positive cases 2020 SARS-CoV2 IgG ELISA IL-1001 IVD Use for Dynex Analyzer 2020 13. DiaSorin. LIAISON® SARS-CoV-2 S1/S2 IgG 2020 CoV-2 Seroconversion in Humans: A Detailed Protocol for a Serological Assay Current protocols in microbiology Predicting detection limits of enzyme-linked 304 ELISA) and bioanalytical techniques in general Clinical 307 Performance of Two SARS-CoV-2 Serologic Assays Clinical 310 Performance of the Roche SARS-CoV-2 Serologic Assay Comparison and development of two different solid phase 313 chemiluminescence ELISA or the determination of albumin in urine