key: cord-0933411-6zfpcm4j
authors: Gittelman, R. M.; Lavezzo, E.; Snyder, T. M.; Zahid, H. J.; Elyanow, R.; Dalai, S.; Kirsch, I.; Baldo, L.; Manuto, L.; Franchin, E.; Del Vecchio, C.; Simeoni, F.; Bordini, J.; Lore, N. I.; Lazarevic, D.; Cirillo, D. M.; Ghia, P.; Toppo, S.; Carlson, J. M.; Robins, H. S.; Tonon, G.; Crisanti, A.
title: Diagnosis and Tracking of Past SARS-CoV-2 Infection in a Large Study of Vo', Italy Through T-Cell Receptor Sequencing
date: 2020-11-12
journal: nan
DOI: 10.1101/2020.11.09.20228023
sha: e3cadfd3a12754ce7d3c8fbaa7447310a8a49eb5
doc_id: 933411
cord_uid: 6zfpcm4j

Measuring the adaptive immune response after SARS-CoV-2 infection may improve our understanding of COVID-19 exposure and potential future protection or immunity. We analyzed T-cell and antibody signatures in a large population study of over 2,200 individuals from the municipality of Vo', Italy, including 70 PCR-confirmed COVID cases (24 asymptomatic, 37 symptomatic, 9 hospitalized). Blood samples taken 60 days after PCR diagnosis demonstrated 97% (68/70) of the latter subjects had a positive T-cell test result, higher than an antibody serology assay (77%; 54/70 of subjects) performed on the same samples. The depth and breadth of the T-cell response was associated with disease severity, with symptomatic and hospitalized COVID cases having significantly higher response than asymptomatic cases. In contrast, antibody levels at this convalescent time point were less informative as they did not correlate with disease severity. 45 additional suspected infections were identified based on T-cell response from the 2,220 subjects without confirmatory PCR tests. Among these, notably, subjects who reported symptoms or had household exposure to a PCR-confirmed infection presented a higher T-cell test positive rate. Taken together, these results establish that T cells are a sensitive, reliable and persistent measure of past SARS-CoV-2 infection.

COVID-19 presents with a wide range in severity, ranging from asymptomatic infection to severe illness and death. To date, it is unknown whether, and for how long, prior infection with SARS-CoV-2 provides immunity against future re-infection, nor is it known how the severity of disease might influence long-term immunity (Poland 2020) . Direct, quantitative measures of the adaptive immune response to SARS-CoV-2 infection, particularly in longitudinal samples following recovery, may offer insights into immunity.

Following the first reported COVID-19 death in the municipality of Vo', Italy and subsequent lockdown of the entire municipality, a large study was undertaken to screen and follow the majority of the residents in that area (Lavezzo 2020) . Two consecutive time points of PCR-based diagnostic tests for SARS-CoV-2 were performed on nearly 2,900 people 1-2 weeks apart, identifying 82 people who were positive for the virus. Approximately 60 days after the second PCR time point, blood samples were collected from the majority of study participants and quantitative assessment of both SARS-CoV-2 specific T cells and IgG antibody titers was performed. As the initial burst of virus-specific effector T cells and secreted antibodies from plasmablasts are likely to have subsided two months post diagnosis, this convalescent time point is appropriate to assess longer term adaptive immune memory.

T-cell receptor (TCR) sequencing was performed on blood samples from 2,290 study subjects using ImmunoSEQ®. Prior research (Snyder 2020) has identified thousands of shared public TCR sequences that are associated with SARS-CoV-2 infection and can be used diagnostically to determine recent or past infection. Using these public sequences, an overall T-cell test result (positive/negative) as well as the clonal depth (relative proportion of T cells that are SARS-CoV-2 specific) and clonal breadth (fraction of all unique TCR DNA clones that are SARS-CoV-2 specific) of response were calculated. Antibody response (result and titer) was measured at the same timepoint for these subjects using a commercial antibody serology test. Of the 2,290 subjects, 70 had a confirmed diagnosis of SARS-CoV-2 from a prior positive PCR test, and the rest remained PCR negative at both surveys. Twenty-four of these 70 individuals reported no symptoms, 37 reported symptoms but did not require hospital admission, and 9 seriously ill patients were hospitalized. Across subjects with confirmed infection, we found that 97% (68 of 70) had detectable SARS-CoV-2 specific T cells and a positive T-cell test result, while only 77% (54 of 70) had a positive serology test result. Notably, the 23% of antibody negative convalescent subjects were distributed relatively evenly between asymptomatic and symptomatic subjects.

Since almost all subjects with prior confirmed infection had readily detectable SARS-CoV-2 specific T cells, we assessed both the depth and breadth of the T-cell response in these subjects. Asymptomatic subjects demonstrated a T-cell response at 60 or more days following confirmed infection that was significantly lower in depth (p=0.002; Mann-Whitney-U test) and also showed trends of less breadth (p=0.16; Mann-Whitney-U test) compared to those with more severe disease ( Figure 1A , B). These results are consistent with other recent findings that the magnitude of T-cell response is higher in symptomatic subjects and these differences may persist for at least 6 months (Zuo 2020), suggesting, as potential hypotheses, that differences in viral load or viral persistence during acute infection correlates both with symptoms and the depth of the T-cell response. In contrast, although a high dynamic range was observed across antibody titers 60+ days post infection, these titers did not correlate with severity of symptoms ( Figure 1C ). Other studies (including Isho 2020, Iyer 2020, Long 2020, Seow 2020) have identified potential differences in antibody levels associated with severity, particularly during the acute phase of illness, but these signals decline with time and seroreversion is sometimes observed. The results here suggest that antibody signals are overall less informative for measuring prior immune response, being highly variable in the months following infection.

The comprehensive survey across the municipality of Vo' also allowed us to explore in detail the general population who tested negative by PCR. Notably, we identified a total of 45 (2.0%) of these 2,220 PCR-negative subjects with a positive T-cell test result, with a level of T-cell response similar to many PCR-confirmed infections ( Figure 1A , 1B and 2A). Within the sampled population of PCR-negative subjects, 249 reported having some symptoms; 8.0% (20 of these subjects) had a positive T-cell test. We next explored whether individuals residing in the same household with an individual with SARS-CoV-2, but were nonetheless PCR negative, demonstrated a T-cell response. 25% (15 out of 60) of PCR-negative individuals with household exposure were T-cell test positives. Overall, positivity rates for the T-cell response were 1.1% among individuals with no reported symptoms or household exposure compared to 55% among individuals who reported both symptoms and household exposure ( Figure 2B ). Importantly, of the 20 subjects that were symptomatic and had detectable T-cell responses, six were symptomatic prior to the PCR surveys, indicating they may have resolved infection by the time of initial survey. 11 of the 20 subjects were symptomatic during the time of survey and may have tested falsely negative by PCR, while the remainder only reported symptoms after the PCR surveys. Antibody serology results were concordant with the T-cell signature for a majority of the individuals with self-reported symptoms or household exposure (Supporting Figure S1 , Supporting Table S1 ). However, serology had a higher percentage of positivity (2.0% compared to 1.1%) in subjects without symptoms or household exposure, which may indicate differences in performance for the two tests. Additional serologic testing and analysis of contact tracing information is currently being performed and will be reported soon.

Taken together, these data indicate that surveying the T-cell response to SARS-CoV-2 can identify cases of prior SARS-CoV-2 infection missed by PCR sampling at high sensitivity and specificity several months after potential infection.

In this study we present evidence, gathered in a carefully selected population, that the T-cell response is a highly-sensitive indicator of prior SARS-CoV-2 infection. We found that T-cell responses were present at least two months after infection, even in subjects who were asymptomatic, potentially confirming prior infection in individuals who may have been exposed but did not have a confirmatory PCR test.

The findings that T-cell immunity appears more sensitive than the antibody response in detecting SARS-CoV-2 infection have important conceptual and diagnostic implications, and yet are not entirely unexpected. Previous experiences with the related Middle East Respiratory Syndrome (MERS) and SARS-CoV-1 infections demonstrated that coronavirus-specific T cells have long term persistence and contribute to protection even in individuals without seroconversion (Channappanavar 2014 , Zhao 2017 . Recent evidence suggests that a similar pattern is present during SARS-CoV-2 infection (Gallais 2020, Thieme 2020).

T-cell response also correlated with disease symptoms and severity, suggesting that the engagement of T cells is finely tuned to viral burden as well as persistence. This correlation is intriguing, as it has not been established whether the T-cell response is exclusively beneficial, or . 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) preprint

The copyright holder for this this version posted November 12, 2020. ;  whether it might also contribute to immunopathology (Altmann 2020). Cases of re-infection have been described (Mulder 2020, Tillett 2020) but as they accumulate and are studied in detail, potential differences in the rates, susceptibility, or severity of re-infection may be explained by the nature of the adaptive immune response, including disease-specific memory T cells, as a measurable and reliable correlate of protection. Additional studies are imperative to define the role of T cells in SARS-CoV-2 infection and immunity.

Clinical data for this cohort (as described in Lavezzo 2020) is available at https://github.com/ncov-ic/SEIR_Covid_Vo. T-cell repertoire profiles are available as part of the ImmuneCODE data resource (Nolan 2020 

This report extends results for the Vo', Italy cohort initially described in Lavezzo 2020. Upon the detection of SARS-CoV-2 in a deceased resident of Vo' on 21 February 2020, an epidemiological study was conducted to investigate the prevalence of SARS-CoV-2 infection in the municipality. Sampling for viral PCR testing was performed on the majority of the population immediately after the detection of the first cases (21-29 February 2020) and again at the end of a 2-week lockdown (7 March 2020). Follow-up blood samples were collected 56 days later in early May for antibody serology and T-cell testing. Antibody response was measured using a commercial serology test of IgG levels against spike protein, to be described elsewhere.

In addition to biospecimen collection, clinical data was collected for each study participant including the results of SARS-CoV-2 testing, demographics, health records, and residence and contact network information. The definition of symptomatic used in this study is a participant who required hospitalization and/or reported fever (yes/no or a temperature above 37 °C) and/or cough and/or at least two of the following symptoms: sore throat, headache, diarrhoea, vomit, asthenia, muscle pain, joint pain, loss of taste or smell, or shortness of breath. Symptomatic subjects who reported hospitalization are split out separately as "hospitalized" in the disease severity analyses.

Genomic DNA was extracted from frozen, plasma-depleted blood samples using the Qiagen DNeasy Blood Extraction Kit (Qiagen). As much as 18 g of input DNA was then used to perform immunosequencing of the CDR3 regions of TCR chains using the ImmunoSEQ Assay. Briefly, input DNA was amplified in a bias-controlled multiplex PCR, followed by high-throughput sequencing. Sequences were collapsed and filtered to identify and quantitate the absolute abundance of each unique TCRβ CDR3 region for further analysis as previously described (Robins 2009, Robins 2012, Carlson 2013). In order to quantify the proportion of T cells out of total nucleated cells input for sequencing, or T cell fraction, a panel of reference genes present in all nucleated cells was amplified simultaneously (Pruessmann 2020).

Classification of prior infection with SARS-CoV-2 as well as the clonal depth and breadth of Tcell response were calculated using a method similar to prior work . Briefly, T-cell receptor repertoires from 784 unique cases of RT-PCR confirmed SARS-CoV-2 infection and 2,448 healthy controls collected before 2020 were compared by one-tailed Fisher's exact tests to identify 4,469 public TCRβ sequences ("enhanced sequences") significantly enriched in SARS-CoV-2 positive samples. (For clarity, all training data to identify the enhanced sequences for SARS-CoV-2 infection came from multiple other study cohorts and not the population being analyzed here.) The enhanced sequences were used to develop a classifier predicting current or past infection with SARS-CoV-2 using a simple two feature logistic regression with dependent variables E and N, where E is the number of unique TCRβ DNA sequences that encode an enhanced sequence and N is the total number of unique TCRβ DNA sequences in that subject. Application of this initial clinical classifier to this study demonstrated the high sensitivity (97%) reported above.

We have since developed a method to improve specificity near the decision boundary of the logistic regression by filtering enhanced sequences that may be potential false positives. Specifically, T-cell receptors that are likely associated with CMV or with multiple antigens in different HLA backgrounds and thus not truly diagnostic of SARS-CoV-2 infection are identified by Fisher's Exact testing on TCRβ repertoires of ~2,000 healthy controls with available HLA genotyping and CMV serotyping data. From this list of ~1.8M sequences, the 182 sequences that were also identified as SARS-CoV-2 enhanced sequences were removed, leaving 4,287 enhanced sequences. The two-feature logistic regression classifier was refitted to the original training data using this pruned enhanced sequence list, and a decision boundary representing 99.8% specificity on 1,657 controls was used to define the test-positive threshold used in the present study. The pruned list of enhanced sequences was also used to calculate the clonal depth and breadth using the same formulae as in .

. 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) preprint

The copyright holder for this this version posted November 12, 2020. ; https://doi.org/10.1101/2020.11.09.20228023 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 12, 2020. ; Figure 2 : T-cell signature across entire cohort, included exposed cases without confirmed diagnosis. 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) preprint

The copyright holder for this this version posted November 12, 2020. ; https://doi.org/10.1101/2020.11.09.20228023 doi: medRxiv preprint Supporting Figure S1 : Percentage of antibody test positives in individuals without a confirmed PCR+ result. Positive antibody test results were observed more often in exposed and/or symptomatic individuals.

Antibody signal in PCR-donors . 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) preprint

The copyright holder for this this version posted November 12, 2020. ; PCR negative, exposed or symptomatic

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) preprint

The copyright holder for this this version posted November 12, 2020. ; https://doi.org/10.1101/2020.11.09.20228023 doi: medRxiv preprint

SARS-CoV-2 T cell immunity: Specificity, function, durability, and role in protection

Virus-Specific Memory CD8 T Cells Provide Substantial Protection from Lethal Severe Acute Respiratory Syndrome Coronavirus Infection

Persistence of serum and saliva antibody responses to SARS

Imperial College COVID-19 Response Team

Clinical and immunological assessment of asymptomatic SARS-CoV-2 infections

preprint. A large-scale database of T-cell receptor beta (TCRb) sequences and binding associations from natural and synthetic exposure to SARS-CoV-2

SARS-CoV-2 immunity: review and applications to phase 3 vaccine candidates

medRxiv preprint. The SARS-CoV-2 T-cell immunity is directed against the spike

Genomic evidence for reinfection with SARS-CoV-2: a case study

We thank the population of Vo' for volunteering en masse to participate in this study. We would like to thank Damon May for helpful discussions, and Beryl Crossley and Mitch Pesesky for assistance reviewing the data.

This work was supported by the Veneto region and was jointly funded by the UK Medical Research Council (MRC; grant MR/R015600/1), the UK Department for International Development (DFID) under the MRC/DFID Concordat agreement, the Abdul Latif Jameel Foundation and is also part of the EDCTP2 programme supported by the European Union. Support was also provided by the Fondazione Umberto Veronesi, Misura Ricerca Covid 19, year 2020.

RMG, TMS, RE, SD, IK, LB have employment and equity ownership with Adaptive Biotechnologies. HSR has employment, equity ownership, patents, and royalties with Adaptive Biotechnologies. HJZ and JMC have employment and equity ownership with Microsoft.