key: cord-0319086-fcfdqe00 authors: Cole, M.; Duffy, E. R.; Osterland, J. N.; Gawel, S.; Ye, L.; de la Cena, K.; Ragan, E. J.; Weber, S. E.; Schechter-Perkins, E. M.; Bouton, T. C.; Jacobson, K. R.; Andry, C.; Kataria, Y. title: Reevaluation of Seroprevalence using a Semi-quantitative Anti-spike IgG in Health Care workers at an Academic Medical Center in Boston, Massachusetts date: 2022-01-21 journal: nan DOI: 10.1101/2022.01.20.22269543 sha: d489ded30bc7b96daa276068e676a839313c9c17 doc_id: 319086 cord_uid: fcfdqe00 Background: Accurate measurement of antibodies is a necessary tool for assessing exposure to SARS-CoV-2 and facilitating understanding of the role of antibodies in immunity. Most assays are qualitative in nature and employ a threshold to determine presence of antibodies. Semi-quantitative assays are now available. Here we evaluate the semi-quantitative SARS-CoV-2 IgG II (anti-spike (S)) assay. We aim to reassess the seroprevalence using anti- S assay and subsequently compare it to the previously measured IgG (anti-nucleoprotein (N)) in health care workers at an academic medical center in Boston. Methods: 1743 serum samples from HCWs at Boston Medical Center were analyzed for SARS-CoV-2 anti-S IgG and IgM using the Abbott SARS-CoV-2 IgG II andAbbott AdviseDxSARS-CoV-2 IgM assay, respectively. Precision, linearity, positive and negative concordance with prior RT-PCR test were evaluated for anti-S IgG. Seroprevalence and its association with demographics variables was also assessed. Results: Linearity and precision results were clinically acceptable. The positive and negative concordance for anti-S IgG with RT-PCR was 88.2% (95% CI: 79.4% - 94.2%) and 97.43% (95% CI: 95.2% - 98.8%), respectively. Overall, 126 (7.2%) of 1,743 participants were positive by anti-S IgG. Among the 1302 participants with no prior RT-PCR, 40 (3.1%) were positive for anti-S IgG antibody. The original agreement in this population with the qualitative, anti-N IgG assay was 70.6%. Upon optimizing the threshold from 1.4 to 0.49 S/CO of the anti-N IgG assay, the positive agreement of the assay increases to 84.7%. Conclusion: The anti-S IgG assay demonstrated reproducible and reliable measurements. This study highlights the presence of asymptomatic transmission among individuals with no prior history of positive RT-PCR. It also highlights the need for optimizing thresholds of the qualitative SARS-CoV-2 IgG assay for better agreement between assays by the same vendor. assay SARS-CoV-2 IgG (anti-nucleoprotein (N)) assay (11) . Seroprevalence studies can assist 98 in detecting asymptomatic and symptomatic infection and provide a cumulative prevalence 99 estimate. Abbott Laboratories has now been issued an EUA for its anti-spike quantitative SARS-Baseline samples were obtained from an ongoing longitudinal cohort assessing 110 serological status among HCW's at Boston Medical Center (BMC). In brief, participants were The analytical measuring range of anti-S IgG is 22.0 -25,000 AU/mL. The suggested 169 manufacturer dilution is 1:2, extending the measuring upper limit to 50,000 AU/mL. Dilution studies were carried out using two participants that had elevated serum samples. Although the EUA approved upper linearity is 25,000 AU/mL, reagent was received prior to EUA 172 approval and was reported to be 50,000 AU/mL at the time of these analyses. These samples CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 excluded from analysis. PCR) as the gold standard among participants with a prior RT-PCR test result. Comparison The data suggests that the difference between expected and measured is within the 95% CI. Table 1 shows demographics in relation to anti-S IgG status (n = 1,743). Overall, 126 of 1,743 225 (7.23%) participants were anti-S IgG positive. Participants who were female, Hispanic, or Black 226 were more likely to be seropositive, but these findings did not reach statistical significance. Obese participants were two times more likely to be seropositive for anti-S IgG [RR: 2.04 (95% CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted January 21, 2022. Overall agreement between the two assays in this population was found to be 97.5% (95% CI: 260 96.7% -98.2%). Of note, the positive agreement was 70.6% (95% CI: 61.8% -78.4%), whereas 261 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) statistically different (p<0.001). A total of 37 participants had detectable anti-S IgG but were negative by anti-N IgG (Table 265 3a). Of these, 11 participants had a positive RT-PCR, 3 participants had a negative RT-PCR, This study reevaluates the seroprevalence in HCWs using a semi-quantitative anti-S IgG assay. It also compares the findings to the anti-N IgG in the same population. We report an overall 286 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 21, 2022. The anti-S IgG shows acceptable intra-and inter-day precision. The precision for the negative 295 control is expected to be high due to a lower numerical value and small variations observed 296 artificially increase the CV. Our results depict acceptable linearity up to 50,000 AU/mL which 297 exceeds upper limit stated in the package insert. It is currently unclear what titer levels confer 298 SARS-CoV-2 immunity but our data suggest that such elevated levels are not rare among the 299 participants. Our results indicate less than ideal positive concordance for anti-S IgG vs. RT-PCR testing 302 which is inconsistent with the manufacturer claims. Further investigation of the discordant 303 samples (n=10) unveil that these participants did not have detectable anti-N IgG, or IgM 304 antibody. These findings can possibly be explained by mild symptoms and/or disease. Literature 305 suggests that the intensity and longevity of antibody response is associated with disease 306 severity (17-19). The negative concordance of the SARS-CoV-2 assay was robust and 307 consistent with the manufacturer claims. Neither SARS-CoV-2 anti-N nor anti-S IgG assays find an association with sex, race, smoking 310 status. However, the seroprevalence determined by anti-S IgG assay was higher (7.2%) relative 311 to the anti-N IgG assay (5.5%) (8). The observed seroprevalence at a regional Boston area 312 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101/2022.01.20.22269543 doi: medRxiv preprint a seroprevalence range between 14% to 25% (21). Seroprevalence can vary between sampling 314 times and geographic regions. Among participants that had no prior RT-PCR test, 40 315 participants had detectable anti-S IgG antibody and 23 participants had detectable anti-N IgG 316 antibody. As such, 17 additional individuals were identified to have IgG antibodies. This 317 suggests higher sensitivity of the anti-S IgG assay which is corroborated by others (22). Alternatively, it may be attributable to a shorter half-life of anti-N IgG antibody (7). The qualitative nature of the anti-N IgG assay enabled us to optimize the pre-set threshold. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10. 1101 /2022 independently suggested the same optimized threshold to increase the sensitivity of the assay 339 (28). Of interest, the same cut-off is currently used in Europe and provides external validation of 340 our findings. Taken together, our data lend further support for optimizing the assay threshold to 341 achieve better performance characteristics. These differences between serological assays are also observed between vendors as well. In a 344 head-to-head comparison of five semi-quantitative SARS-CoV-2 IgG assays found that the 345 results from assays are not interchangeable despite good correlation to neutralizing antibody for 346 some of them (10). This has been supported by others (9, 29, 30). Collectively, this highlights 347 the need for harmonization between all serological assays. This is of paramount importance as 348 it will enable us to draw meaningful conclusions about correlation of immunity and being better 349 equipped to overcoming the SARS-CoV-2 pandemic. The present study benefited from a large sample size. Certain limitations are acknowledged. The gold standard for determining protective antibody is a virus neutralization test which we 353 were unable to compare to the anti-N & anti-S assay as it requires live pathogen and a biosafety 354 level 3 laboratory. The study is limited by a cross sectional study design which may under-or 355 overestimate the seroprevalence. We were unable to correlate days since RT-PCR result or 356 viral load with quantitative antibody levels and it may provide insight about antibody kinetics 357 (31). The samples were obtained from a single timepoint preventing characterization of antibody 358 kinetics on performance characteristics; however, a subset of participants is being prospectively 359 followed in three-month intervals which will enable future analysis. Lastly, the findings are not 360 generalizable to the larger community and is limited to hospitals. In conclusion, the Abbott SARS-CoV-2 anti-S IgG assay demonstrate acceptable performance 363 characteristics. The study highlights the presence of infection among participants with no RT- . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 21, 2022. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 21, 2022. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 21, 2022. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 21, 2022. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101/2022.01.20.22269543 doi: medRxiv preprint Fisher's Exact Test P value: <0.0001 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101/2022.01.20.22269543 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101/2022.01.20.22269543 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10. 1101 /2022 Hispanic/LatinX 0 . .