key: cord-318789-ylxh8vi2
authors: Byrne, R. L.; Kay, G. A.; Kontogianni, K.; Brown, L.; Collins, A. M.; Cuevas, L. E.; Ferreira, D.; Fraser, A. J.; Garrod, G.; Hill, H.; Menzies, S.; Mitsi, E.; Owen, S. I.; Williams, C. T.; Hyder-Wright, A.; Adams, E. R.; Cubas-Atienzar, A. I.
title: Saliva offers a sensitive, specific and non-invasive alternative to upper respiratory swabs for SARS-CoV-2 diagnosis.
date: 2020-07-11
journal: nan
DOI: 10.1101/2020.07.09.20149534
sha: 
doc_id: 318789
cord_uid: ylxh8vi2

RT-qPCR utilising upper respiratory swabs are the diagnostic gold standard for SARS-CoV-2 despite reported low sensitivity and limited scale up due to global shortages. Saliva is a non-invasive, equipment independent alternative to swabs. We collected 145 paired saliva and nasal/throat (NT) swabs at diagnosis (day 0) and repeated on day 2 and day 7 dependent on inpatient care and day 28 for study follow up. Laboratory cultured virus was used to determine the analytical sensitivity of spiked saliva and swabs containing amies preservation media. Self-collected saliva samples were found to be consistent, and in some cases superior when compared to healthcare worker collected NT swabs from COVID-19 suspected participants. We report for the first time the analytical limit of detection of 10-2 and 100 pfu/ml for saliva and swabs respectively. Saliva is a easily self-collected, highly sensitive specimen for the detection of SARS-CoV-2.

This study was embedded into a prospective study evaluating the performance of multiple diagnostic tests for COVID-19 (Facilitating A SARS Cov-2 Test for Rapid Triage (FASTER). Adults with signs and symptoms of suspected COVID-19 attending the Royal Liverpool University (RLUH) and Aintree University Hospitals (AUH) in Liverpool, UK were asked to provide specimens within 24 hours of informed consent given on the day of presentation (D0 ±1 day), figure 1 outlines the studies inclusion and exclusion criteria. 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 (which was not certified by peer review)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 July 11, 2020. . https://doi.org/10.1101/2020.07.09.20149534 doi: medRxiv preprint

NT swabs (Copan, Italy) were taken by a research nurse delegated to the study.

Participants were asked to collect saliva in their mouth and to gently spit it by aid of a funnel in a sterile cryotube collection tube (SARSTEDT, Germany). Saliva volumes were checked by eye and a minimum of approximately 200 microlitres were collected, samples with less volume were excluded from analysis. Samples were transported on ice within 3 hours of collection to the Liverpool School of Tropical Medicine (LSTM) laboratories for processing. RNA extraction was performed immediately on swab samples, while saliva samples were stored at -80 o C until processing.

Fourteen NT swabs were stored in 1ml of amies preservation medium (Copan, Italy) and 4ml of saliva were collected from a confirmed SARS-CoV-2 negative volunteer.

A SARS-CoV-2 isolate (REMRQ0001/human/2020/Liverpool) was propagated in Vero E6 cells (C1008; African green monkey kidney cells), as previously described (11). A serial dilution series of SARS-CoV-2 ranging from 10 6 to 10 -6 plaque forming units per ml (pfu/ml) was used to spike 140µl saliva and swab samples. The LOD was determined by the lowest concentration for which all three PCR replicates amplified.

Viral RNA was extracted using the QIAamp Viral RNA Mini Kit (Qiagen, Germany) following the manufacturer's instructions with an internal extraction control incorporated at the lysis stage (Genesig, UK). Non-spiked saliva and NT swab . 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 July 11, 2020. . https://doi.org/10.1101/2020.07.09.20149534 doi: medRxiv preprint samples from confirmed negative volunteers were included for both matrices as negative extraction controls. Once extracted, samples were taken immediately for downstream application and stored on ice during PCR setup.

For SARS-CoV-2 RT-qPCR detection, 8µl of extracted RNA was tested using the Genesig® Real-Time Coronavirus COVID-19 PCR assay (Genesig, UK) in a RGQ 6000 thermocycler (Qiagen, Germany). Samples for analytical sensitivity were tested in triplicate, including the negative controls. Samples were classified as RT-qPCR positive if both the internal extraction and the SARS-CoV-2 probes were detected at <40Ct. Virus copies/ml were quantified using the manufacturer's positive control (1.67

x 10 5 copies/µl) as a reference.

The analytical sensitivity in spiked samples indicated the LOD for saliva was 10 -2 copies/ml and for the NT swabs 10 0 copies/ml. However, amplification of one or more replicates was recorded for saliva and NT swabs to concentrations 10 -6 (copies/ml) and 10 -4 (copies/ml), respectively.

One hundred and ten adults were recruited between April and June 2020. Of these, 61 were female, 49 males. Most participants were hospitalised, with only 21 (19%) discharged home directly from A&E.

. 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 July 11, 2020. . https://doi.org/10.1101/2020.07.09.20149534 doi: medRxiv preprint

In total, 145 paired saliva and NT swabs were collected. Most paired samples were taken at the time of diagnosis (n=110), with follow up samples on D2 (n=14) and D7 (n=6) were only available from positive hospitalised participants. Only 15 participants had returned on D28 at the time of publishing the results.

The proportions of RT-qPCR positive samples for both specimens are shown in Table 1 . Overall, 19 saliva and 19 NT swabs (11.6%) of 145 paired samples were positive. As expected, the proportion positive varied with time, with both saliva and NT swabs being more likely to be positive on D2 and D7 than on D28, although numbers are too small to reach statistically significant differences. The viral loads for saliva and NT swabs ranged from 36 to 5.4 x 10 7 copies/ml for both on enrolment.

Overall VLs were similar, with a good agreement between the samples, as shown in 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 July 11, 2020. . https://doi.org/10.1101/2020.07.09.20149534 doi: medRxiv preprint qPCR. Of these, five were positive in both samples, one was NT swab-positive but saliva negative and 3 were saliva-positive but NT swab-negative. For one participant, the saliva sample was consistently positive at D0, D2, D7, and D28, whereas the NT swab was positive at day D0, D2 and D7 but negative at D28. All discordant samples that had failed to amplify in saliva or NT swab, the positive sample had a viral load ≤10 1 (copies/ml) and Ct values >36 indicating viral titres were low. . 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 July 11, 2020. . https://doi.org/10.1101/2020.07.09.20149534 doi: medRxiv preprint

Here we report for the first time the analytical sensitivity of saliva for the detection of SARS-CoV-2 compared to NT swab samples. In spiked saliva samples, SARS-CoV-2 can be detected with greater sensitivity in saliva compared to NT swabs by 100fold copies/ml, presumably due to the dilution factor of the amies transport medium.

We also report the agreement between paired saliva and NT swabs in clinical samples of participants attending hospital settings. Overall, saliva had a good agreement to NT swabs during hospitalisation (D0, D2 and D7) and after recovery (D28). There were minor disagreements between the samples, which were not statistically significant, and these disagreements were bi-directional, with slightly more samples being NT swab positive on enrolment, but more saliva samples being positive during the follow up period. All saliva-positive, NT swab-negative samples in D2 (n=2) and D28 (n=1) reported low viral titre (<10 1 copies/ml) and are likely due to similar limitations outlined in the analytical sensitivity, a greater dilution factor of amies. It is important to note that these were from participants that had tested positive on D0 and thus likely to be true positive and support saliva to be more consistent for temporal sampling. This is in agreement with Wyllie et al., (2020) who reported that saliva had a greater sensitivity and was more likely to be constantly positive throughout the course of infection on a subset of COVID-19 hospitalised participants (n=29).

All saliva-negative but NT swabs-positive were also low viral titre and may have resulted from the different processing time of the two specimens. As NT swabs underwent RNA extraction immediately, saliva samples were stored at -80 o C prior to RNA extraction. Freeze-thawing can have a significant impact on the quality of RNA . 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 July 11, 2020. There will be numerous applications also in the UK for home sampling and the screening of children, who have previously been shown to reject swabbing. As well as in research studies that require repeat sampling and due to the non-invasive nature of saliva will likely promote compliance by participants. . 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 July 11, 2020. . https://doi.org/10.1101/2020.07.09.20149534 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 July 11, 2020. . https://doi.org/10.1101/2020.07.09.20149534 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 July 11, 2020. . https://doi.org/10.1101/2020.07.09.20149534 doi: medRxiv preprint

World Health Organization. Laboratory testing strategy recommendations for COVID-19

Interim Guidelines for Clinical Specimens for COVID-19 | CDC

Temporal profiles of viral load in posterior oropharyngeal saliva samples and serum antibody responses during infection by SARS-CoV-2: an observational cohort study

Molecular and serological investigation of 2019-nCoV infected participants: implication of multiple shedding routes. Emerg Microbes Infect

Saliva is more sensitive for SARS-CoV-2 detection in COVID-19 participants than nasopharyngeal swabs

We are grateful to all the participants recruited to the FASTER clinical research project, for taking the time in this pandemic situation to be involved in research. We also thank the LUHFT and NIHR research nurses and LSTM team who assisted with the sample collection and processing. We thank Ian Pattinson for assistance with