key: cord-0876762-tveeq4fj
authors: Özçürümez, Mustafa K.; Ambrosch, Andreas; Frey, Oliver; Haselmann, Verena; Holdenrieder, Stefan; Kiehntopf, Michael; Neumaier, Michael; Walter, Michael; Wenzel, Folker; Wölfel, Roman; Renz, Harald
title: SARS-CoV-2 Antibody Testing – Questions to be asked
date: 2020-05-29
journal: J Allergy Clin Immunol
DOI: 10.1016/j.jaci.2020.05.020
sha: 9c5d408cbcc983655a26c45865e9c9216399a45e
doc_id: 876762
cord_uid: tveeq4fj

ABSTRACT SARS-CoV-2 infection and development of COVID-19 disease presents a major healthcare challenge of global dimensions. Laboratory diagnostics of infected patients, and the assessment of immunity against the SARS-CoV-2 virus presents a major cornerstone in handling the pandemic. Currently there is an increase in demand of antibody testing and a large number of tests are already marketed or in the late stage of development. However, the interpretation of test results depends on many variables and factors, including sensitivity, specificity, potential cross-reactivity and cross-protectivity; the diagnostic value of antibodies of different isotypes, the use of antibody testing in identification of acutely ill patients or in epidemiological settings. In this article the recently established COVID-19 Task Force of the German Society for Clinical Chemistry and Laboratory Medicine (DGKL) addresses these issues based on the currently available datasets in this rapidly moving field.

4710 words 48 patients, and the assessment of immunity against the SARS-CoV-2 virus presents a 52 major cornerstone in handling the pandemic. Currently there is an increase in 53 demand of antibody testing and a large number of tests are already marketed or in 54 the late stage of development. However, the interpretation of test results depends on 55 many variables and factors, including sensitivity, specificity, potential cross-reactivity 56 and cross-protectivity; the diagnostic value of antibodies of different isotypes, the use 57 of antibody testing in identification of acutely ill patients or in epidemiological settings. 58

In this article the recently established COVID-19 Task Force of the German Society 59

for Clinical Chemistry and Laboratory Medicine (DGKL) addresses these issues 60 based on the currently available datasets in this rapidly moving field. 61

Antibody response; COVID-19; diagnostic pathway; external quality assurance; 63 immunity; immunoassay; neutralization assay; respiratory infections; serologic 64 analysis; severe acute respiratory syndrome coronavirus 2. diagnostics were necessary and similar challenges existed with regard to the 85 evaluation of test results 1 . A major difference to that time is the strong political and 86 economic pressure to insist on the most reliable high-throughput diagnostics. There 87 is an urgent need for the development of appropriate laboratory tests to identify 88 infected patients, follow the course of viral shedding and clearance and to assess 89 immunity against SARS-CoV-2. Laboratory testing is built on two different pillars: on 90 the one side, the detection and measurement of viral RNA, and on the other side 91 measuring antibodies of various isotypes against SARS-CoV-2 components, 92 reflecting the host immune response. Although antibodies are developing quite early 93 during the course of the disease, the serological response is not suitable for early 94 detection of infected patients. Furthermore, the clinical and immunological meaning 95 of these antibody responses is unclear, since the many available tests do not 96 necessarily prove protective immunity against the SARS-CoV-2 virus. Furthermore, it 97 can be questioned, whether serological testing can be used as a surrogate marker 98 for viral encounter. In this regard it remains unclear whether oligo-or 99 monosymptomatic cases -which are still the majority of all SARS-CoV-2 infected 100 patients -also develop this type of immune response. In addition, the longevity of the 101 persistence of these antibodies is still not clear. There is increasing interest to use 102 antibody testing to assess the immune status of larger populations and also of the 103 risk population such as healthcare workers and others, in order to help to draw conclusions from drastic measures such as economic and social lockdown, social 105 distancing and other restrictive actions. These key questions require immediate 106 attention, in order to appreciate the strength and weakness of antibody testing 107 against SARS-CoV-2. This article summarizes the currently available knowledge and 108 literature in this extremely rapidly moving area. 109

What are the approved indications to perform a COVID-19 serology? 111

In most patients, antibodies against SARS-CoV-2 become detectable within the first 112 10 days after onset of symptoms of COVID-19. Also the kinetics of the class switch of 113 different isotypes of SARS-CoV-2 specific immunoglobulins is comparable to other 114 coronavirus infections 2-10 . IgM, IgA and IgG antibodies were detectable in some 115 patients as early as day 1 after onset of symptoms. The interquartile ranges of the 116 first antibody detection for IgM and IgA are between day 3 and 6, and for IgG 117 between day 10 and 18. IgA reached a plateau up to day 7, while IgM and IgG 118 continuously increased until day 14 and day 21, respectively 5 . Therefore, serological 119 testing could be useful in several different aspects of COVID-19 11 . 120

First, and perhaps most important, serological testing could supplement standard RT-121 PCR assays for diagnosis of COVID-19 in symptomatic patients. There is 122 accumulating evidence that viral shedding in the upper respiratory system profoundly 123 decreases 7-10 days after infection, leading to negative swab results in at least 30-50 124 percent of COVID-19 cases 6,12-14 . Measurement of SARS-CoV-2-specific antibodies, 125 which begin to be detectable in a significant proportion of patients 5-7 days after 126 infection and later in almost all cases, could help to detect cases with negative RT-since the immunological response triggered by an acute infection like COVID-19 has 129 a certain latency. 130

Second, serological testing is considered to be used to retrospectively determine 131 SARS-CoV-2 infections in people that previously have not tested positive by RT-PCR 132 for whatever reasons. However, the kinetics and the magnitude of the antibody 133 response seems to correlate with the clinical severity of the disease 4,5 . Preliminary 134 data suggest that an yet unknown number of asymptomatic infected and even 135 oligosymptomatic COVID-19 patients do not develop seroconversion 16, 17 . 136

There is a lack of validation data from IVD manufacturers who have systematically 137 examined asymptomatically infected patients. Therefore, it is currently challenging to 138 establish cut-off values that are sensitive enough to determine the prevalence of 139 infection at the population level without running the risk of too high rates of false-140 positive results. Performance data about the Roche antibody assay have been 141 currently released 18 . The assay exhibited no cross-reactivity with 40 endemic HCoV 142 convalescence sera, i.e. it yieled a specificity of 100% (95% confidence interval 143 91.2% to 100%). More striking, among 5272 pre-CoViD-19 sera collected from 144 routine labs (n = 3420) and blood donors (n = 1772) only 10 reactive sera were 145 identified, i.e. a specificity of 99.81% (95% confidence interval 99.65% to 99.91%) 146 was achieved. With increasing knowledge about SARS-CoV-2, the problem of 147 specificity could fade into the background in the future and the use of serology as an 148 epidemiological instrument become the next challenge. 149

Third, and of utmost importance for the healthcare system and political decisions on 150 lock down measures, is the ability of serological testing to establish indicators of 151 protection against (re-) infection with SARS-CoV-2. Indeed, sera from patients with plasma transfer from convalescent COVID-19 patients demonstrate also in vivo 154 effects 4,19-21 . However, the efficacy of this therapy has not yet been confirmed in 155 sufficiently large, controlled studies. Furthermore, no direct conclusion can be drawn 156 about a reliable protective effect of the antibodies individually acquired during an 157 infection. It is therefore conceivable that anti-SARS-CoV-2 antibodies can protect 158 against the virus. However, demonstrating a neutralizing activity of an antibody 159 against a virus requires assays using live or pseudotyped virus, which cannot be 160 performed in a high-throughput fashion. It is necessary to determine the targets of 161 protective antibodies in order to develop simple immunoassays that best reflect virus 162 neutralization. This is especially important since certain target epitopes of antibodies 163 might also enhance virus entry 22 . Therefore, total antibody measurements do not 164 necessarily reflect protection after infection, nor do they indicate the efficacy of a 165 vaccination to ascertain immunity. 166

In a cross-validation of 22 assays (lateral flow tests and ELISAs) to detect IgM and 168

IgG antibodies in COVID-19 patients, a significant number of positive results were 169 also found in historic sera from the pre-COVID-19 era and from non-SARS-CoV-2 170 infections 23,24 resulting in test specificities ranging from 84% to 100% for both 171 isotypes (95% confidence intervals 76% to 91% and 97% to 100%, respectively). The 172

reported specificity of 100% for both, IgG and IgM, was yielded by one of the lateral 173 flow assays, however, especially evident for IgM, sensitivity within the first 10 days 174 after patient reported symptom onset was lower as compared to the other assays. 175 is the main determinant of the positive predictive value (PPV). The recently reported 177 prevalence of COVID-19 in the population 25,26 of 1 % to 4 % will result in a PPV 178 between 25% and 58% assuming a specificity of 97% and between 4% and 15% for 179 76% specificity, respectively, at an artificial sensitivity of 100% in all scenarios It is 180 therefore not possible to infer protection against SARS-CoV-2 from a positive result 181 of an immunoassay (see Figure 1 ). 182 Figure 2 shows an example of PPV / NPV values (y-axis) as a function of prevalence 183 (x-axis) for theoretically assumed test sensitivities/ and specificities from 80% to 184 99.9%, respectively, and for two commercially available SARS-CoV-2 IgG tests with 185 sensitivities of 88.66% and 80% and specificities of 90.63% and 98.5%, respectively. 186

Applying these assay performance figures to testing strategies in the general 187 population, predictive values of 2.2% to 7.9% (PPV) and 99.97% to 99.89% (NPV) or 188 11.4% to 32.6% (PPV) and 99.95 to 99.82% (NPV) can be calculated for a 189 prevalence of 0.24% (Regensburg, Bavaria, Germany) or 0.9% 26 , respectively. 190 Clinical triage for COVID-19 symptoms increases the pre-test probabilities towards 191 48% in hospitalized settings and will raise the PPV for the same tests to 89.73% and 192 98.01% while NPV slightly decreases to 89.65% and 84.33%, respectively. protein as antigen 26-28 and is completed on day 14, in severe cases of ARDS 207 seroconversion seems to occur earlier 4 ; in mild or asymptomatic cases 208 seroconversion may even be absent 26 . 209

The detection of persistent infectivity cannot be conclusively verified by commercially 211 available RT-PCR because it is not possible to distinguish between replicable virus 212 components and inert genome fragments. It is therefore assumed that RT-PCR 213 results lag behind the actual elimination of SARS-CoV-2 in infected individuals. 214

The virological gold standard to prove infectivity is virus isolation in cell culture 13 IgA within the first week of symptoms, followed by IgM and IgG in the second week. 231

As with SARS and MERS, IgM cannot be detected significantly earlier than IgG 8 . 232

Those antibodies which bind specifically to surface structures of SARS-CoV-2, like 233 the S protein, prevent the virus from interacting with its target cell are called 234 neutralizing antibodies. These antibodies play an important role in virus clearance as 235 they have the ability to block viral infection and are assumed to protect patients. 236

Serological tests for SARS-CoV-2 that are intended to confirm such neutralizing 237 antibodies must therefore be robust to the detection of other, non-neutralizing 238 antibodies. Besides interfering factors that also occur in many other assays, such as 239 heterophilic antibodies or human anti-animal antibodies, immunogenic proteins of 240 closely related human coronaviruses can trigger cross-reactive antibodies in the host. 241

This has been known for many decades and led to the earlier categorization of 242 corona viruses into serogroups 31 . Cross-reactivity with serum samples from HCoV 243 patients has been shown for serological SARS-CoV-2 IgA and IgG antibody assays 4 . 244 Therefore, in order to make a valid serological diagnosis of SARS-CoV-2 neutralizing 245 antibodies, it is essential to exclude cross-reactivity by a second confirmatory test. 246

This is even more important when, as in some commercial immunological test 247 systems, the SARS-CoV-2 nuclear protein or parts thereof are used as an antigen.

not have a neutralizing effect on SARS-CoV-2 because the nuclear protein is located 250 inside the virus and is therefore not directly accessible. 251

Widely accepted confirmatory tests, such as the virus neutralization test 252 recommended by the WHO during the SARS outbreak 32 , are labour intensive, 253 resulting in slow sample throughput in diagnostic laboratories. The establishment of 254 highly specific primary screening assays that avoid false positive results and thus the 255 need for further confirmation is therefore an important objective. Surrogate 256 neutralization assays using pseudotyped virus particles that bear the Spike protein of 257 the SARS-CoV-2 virus do not require work inside high containment laboratories and 258 therefore might offer an alternative testing option in the near future 20,33 . 259

Another challenge for the serological detection of SARS-CoV-2 immunity is the 260 possibility of a low antibody response in mildly infected or even asymptomatic 261 COVID-19 cases. Most severe SARS-CoV-2 infections lead to a robust immune 262 response 10 , but on the other hand, PCR-diagnosed mild or asymptomatic infections 263

can cause variable humoral immune responses that might not be detected by 264 serological tests 20,34 or even fall below the detection limit in several patients within a 265 few weeks (Wölfel, unpublished data) . 266

Cross-reactivity and cross-protectivity may be two sides of the same coin in COVID-267 19, too. SARS-CoV-2 is closely related to HCoV-OC43 (another betacoronaviruses), 268 the most prevalent seasonal coronavirus detected among patients under the age of 269 five 16 . It has been hypothesized before that such a pre-existing cross-immunity may 270 confer protection and/or attenuate the severity of COVID-19 35 . Pre-existing cross-271 protective immunity in individuals previously exposed to antigenically related 272 pathogens have already been demonstrated for pandemic influenza A H1N1 in 273 entry of SARS-CoV-2 into the cell in mice 37 , suggesting the possibility of an 275 mechanism analogous to influenza. Finally, it should be mentioned that relatively 276 nonspecific antibodies, such as might be produced by certain vaccination strategies, 277 are suspected of being able to enhance a pathological immune response 22,38 . 278 However, first studies on vaccine antigens based on the RBD subunit of the S protein 279 did not show any evidence of such an antibody dependent enhancement 39 . 280

A growing number of in vitro diagnostic companies are developing SARS-CoV-2-282 specific antibody tests (see https://www.finddx.org/covid-19/pipeline/). In addition to 283 the differences and problems with test performance described above, the different 284 assay techniques also differ in the conclusions that can be drawn from the results. 285 Table 1 gives an overview of assay techniques used in COVID-19 serology. Different 286 antigens (RBD, N, S1) have already been evaluated in various proprietary and 287 commercial ELISA methods 4 . Antigen selection is one of the crucial aspects of assay 288 development, that determines specificity, availability and scalability for mass 289 production. Recombinant proteins are produced either by prokaryotic or eukaryotic 290 expression systems 40 . Prokaryotic systems achieve higher production rates, but the 291 spectrum of suitable antigens is limited due to the lack of posttranslational 292 modification and may also influence their diagnostic performance 41 . Antigen 293 extraction from complete virus lysate is technically less complex, but requires the 294 availability of ultracentrifugation and a BSL3 containment. Raw lysates are of 295 particular interest in the early stages of outbreaks when purified proteins are not yet 296 available. After separation of the protein fractions, virus lysates for Western blotting 297 are used as a viable option for the validation of immunoassays and are also suitable 298 as confirmatory tests. Due to the high safety requirements these approaches for laboratories 42 . 301

The general issue of low PPV demands either robust sensitivities above 99.99% or a 302 2-tier diagnostic process, i.e. positive screening tests have to be confirmed e.g. by 303

Western blot which is a serological standard for many decades. 304

Neutralization assays are the virological reference method for confirmation of 305 neutralizing antibodies. Plaque reduction neutralization tests and also more rapid 306 microneutralization tests have been described for SARS-Cov-2 antibody testing 14,42 307

As all those techniques rely on usage of whole-virus preparations, they are limited to 308 ideally standardization of cut-offs is one of the essential quality criteria that will affect 343 the intended use of COVID-19 serology. This is emphasized by WHO 344 recommending, as of 24 th April 2020 scientific report, to restrict the use of SARS-345

CoV-2 antibody testing to research settings until its diagnostic reliability is proven by With regard to the quality assurance of SARS-CoV-2 antibody tests, there is an 370 urgent need for suitable reference material, for large-scale validation studies 371 involving various available test systems, and for international proficiency testing 372 initiatives.

There are different definitions of "baseline" samples and baseline studies. A blood 376 draw to obtain a baseline serum sample is recommended for contacts of infected 377 persons as early as possible within the incubation period of contact 59,60 . For patients 378 paired samples are necessary for confirmation with the initial (baseline) sample 379 collected in the first week of illness and the second ideally collected 2-4 weeks later 380

(the optimal timing for convalescent sample needs to be established) 59 . 381

In a representative baseline study, a demographically representative cohort is 382 repeatedly tested to determine the rate of spread of the virus. This can be done by 383 serological analysis on blood donors, by studies in particularly affected places 384 ("hotspots") or nationwide in a carefully controlled population-representative study. development, study participants are urgent to collect blood from study participants 397 awaiting such tests in the near future. These tests could become crucial to obtain fully interpretable and unbiased results from these studies. For example, it has 399 recently been proposed to collect samples and data in advance to test the hypothesis 400 that resilience of the elderly during a pandemic can be improved by countering 401 chronic inflammation (inflammaging) and cellular senescence 61 . 402

While this procedure is straightforward within studies, some countries may need 403 special regulations for implementation in the field of health care. At present it is 404 conceivable that biobanks are established with noble intentions but may then be 405 opened for purposes for which prior consent of the patient would have been required. 406

Similarly, this problem could also affect stored sera from employees. At this point, the 407 officials should verify the legitimacy of a proactive blood collection. 408

Following the introduction of PCR methods, it soon became apparent that the 411 demand for test kits far exceeded their availability. A major difference between 412 molecular and serological diagnostics is that the latter can be performed in almost all 413 diagnostic laboratories; usually equipment is readily available. Personal 414 communication with the IVD industry currently estimates a demand only for a single 415 country like Germany between 2,000,000 and 5,000,000 tests per month. The 416 needed capacities may double as it can be assumed that most of the tested persons 417

have to be re-tested within one month. The assumed increase is also triggered by the 418 examination of contacts of persons tested positive in a low prevalence setting. 419

Production capacities of "high double-digit millions" per month" have already been 420 announced by one manufacturer. It therefore remains to be seen whether the 421 forecasts for both demand and availability will be met. To 

Dynamic Characteristics in Patients with COVID-19

Profile of specific antibodies to SARS-CoV-2: The 497 first report

Re: Profile of specific 499 antibodies to SARS-CoV-2: The first report

Antibody responses to

Petherick A. Developing antibody tests for SARS-CoV-2

Negative Rate of RT-PCR Based SARS-CoV-2 Tests by Time Since 508

Estimating false-negative 510 detection rate of SARS-CoV-2 by RT-PCR. Preprint at medRxiv

Virological assessment of hospitalized patients with COVID-2019

First infection by all four non-severe 519 acute respiratory syndrome human coronaviruses takes place during 520 childhood

Different longitudinal 522 patterns of nucleic acid and serology testing results based on disease severity 523 of COVID-19 patients

Ill Patients With COVID-19 With Convalescent Plasma

Neutralizing antibody 531 responses to SARS-CoV-2 in a COVID-19 recovered patient cohort and their 532 implications. Preprint at medRxiv

Treatment with 534 convalescent plasma for COVID-19 patients in Wuhan

Coronavirus Antibody Tests: Can You Trust the Results? New

Test 543 performance evaluation of SARS-CoV-2 serological assays. Preprint at 544 medRxiv

COVID-19 Antibody Seroprevalence in

Spread of SARS-CoV-2 in the Icelandic Population

Correlation of Chest CT 552 and RT-PCR Testing in Coronavirus Disease 2019 (COVID-19) in China: A 553 Report of 1014 Cases

Successful containment of COVID-19 outbreak in a large maternity and 557 perinatal center while continuing clinical service

Antigenicity of the SARS-CoV-2 Spike Glycoprotein

COVID-19: consider cytokine storm syndromes and immunosuppression

Receptor-Binding Domain Elicits a Potent Neutralizing Response Without 597

SARS-CoV-2 Seroconversion in Humans: A Detailed Protocol for a 601

Serological Assay, Antigen Production, and Test Setup

Evaluation of recombinant 604 nucleocapsid and spike proteins for serological diagnosis of novel coronavirus 605 disease 2019 (COVID-19)

Early viral clearance and 612 antibody kinetics of COVID-19 among asymptomatic carriers. Preprint at 613 medRxriv

Antibody Polyreactivity in Health and Disease: Statu Variabilis

Guidelines on COVID-19 in vitro diagnostic tests and their performance

Eur Union 2020/C 122 I/01. 621 46. Vashist SK

During the Public Health Emergency (Revised)

Epub ahead of 630 print] antibody responses with severity of disease

SARS-CoV-2 Reference Material Kit

Gesellschaft zur Förderung der Qualitätssicherung 643 in medizinischen Laboratorien e.V. Novel coronavirus SARS-CoV-2 (2019-644 nCoV) includeding the next INSTAND EQA scheme

D_announcement_EQAS_novel_coronavirus_SARS-CoV-2.pdf 648 53. Laboratory of the Government Chemist Group. COV -SARS-CoV-2 molecular 649 proficiency test

Quality Control for Molecular Diagnostics

First 659 external quality assurance of antibody diagnostic for SARS-new coronavirus

UK NEQAS Pilot Scheme Developments

Application survey SARS-CoV-2 Immunology

World Health Organization. Laboratory testing for 2019 novel coronavirus 669 (2019-nCoV) in suspected human cases, Interim guidance

European Centre for Disease Prevention and Control. Laboratory support for

COVID-19 in the EU/EEA

The 675 preventive strategy for pandemics in the elderly is to collect in advance 676 samples & data to counteract chronic inflammation (inflammaging)

Projecting the 682 transmission dynamics of SARS-CoV-2 through the postpandemic period

Epub ahead of print] 684 64. World Health Organization. Medical Product Alert N°3/2020. Falsified medical 685 products, including in vitro diagnostics, that claim to prevent, detect, treat or 686 cure COVID-19

developed assay detecting IgM and IgG antibodies (total of 14 test systems) in 694 patient sera and controls

VivaChek IgG; H2: 697 UCP IgG; G2: Sure IgG; F2: Premier IgG

Biomedomics IgG; L: Wondito IgG/IgM

VivaChek IgM; H1: UCP IgM; G1: Sure IgM; F1: Premier 700 IgM; E1: Innovita IgM; D1: DeepBlue IgM; C1: Decombio IgM

A1: BioMedomics IgM

NPV (B) values (y-axis) as a function of prevalence (x-704 axis). Gray lines illustrate a theoretically assumed range of test 705 sensitivities/specificities from 80/80%

Two commercially available SARS-CoV-2 IgG tests are shown with (A) specificities of 707 90.63% (blue) and 98.5% (red), and (B) sensitivities of 88.66% (blue) and 80% (red) 708 respectively. PPV for a population-based prevalence of 0

As obvious 710 in (B), even though assay sensitivity is only 80% due to its higher specificity the red 711 line is located above the grey line that indicates prevalence dependent NPV for 712 sensitivities/specificities of 80%