Large parallel screen of saliva and nasopharyngeal swabs in a test center setting proofs utility of saliva as alternate specimen for SARS-CoV-2 detection by RT-PCR

Background A high volume of testing followed by rapid isolation and quarantine measures is critical to the containment of SARS-CoV-2. RT-PCR of nasopharyngeal swabs (NPS) has been established as sensitive gold standard for the detection of SARS-CoV-2 infection. Yet, additional test strategies are in demand to increase and broaden testing opportunities. As one attractive option, saliva has been discussed as an alternative to NPS as its collection is simple, non-invasive, suited for children and amenable for mass- and home-testing. Methods Here, we report on the outcome of a head-to-head comparison of SARS-CoV-2 detection by RT-PCR in saliva and nasopharyngeal swab (NPS) of 1187 adults and children reporting to outpatient test centers and an emergency unit for an initial SARS-CoV-2 screen. Results In total, 252 individuals were tested SARS-CoV-2 positive in either NPS or saliva. SARS-CoV-2 RT-PCR results in the two specimens showed a high agreement (Overall Percent Agreement = 98.0%). Despite lower viral loads in saliva, we observed sensitive detection of SARS-CoV-2 in saliva up to a threshold of Ct 33 in the corresponding NPS (Positive Percent Agreement = 97.7%). In patients with Ct above 33 in NPS, agreement rate dropped but still reaches notable 55.9%. Conclusion The comprehensive parallel analysis of NPS and saliva reported here establishes saliva as a reliable specimen for the detection of SARS-CoV-2 that can be readily added to the diagnostic portfolio to increase and facilitate testing.


Introduction
opting for a voluntary SARS-CoV-2 test at one of five participating 84 test centers were included. Four centers were dedicated test centers for outpatients and one was an 85 emergency care unit. The study population comprised individuals with SARS-CoV-2 related 86 symptoms based on Swiss testing criteria and asymptomatic individuals with relevant exposure to a 87 SARS-CoV-2 index case. Hospitalized patients were not included. Individuals were included without 88 further selection to avoid skewing. Information on symptomatic or asymptomatic status was collected 89 as part of the regular procedure for SARS-COV-2 testing and reporting based on self-evaluation 90 (asymptomatic/mild/strong) by the participants, as they did not see a physician in the test center 91 setting.

92
Ethical approval 93 The Zurich Cantonal Ethics Commission waived the necessity for a formal ethical evaluation based on 94 the Swiss law on research on human subjects, as the collection of saliva in parallel to a scheduled 95 nasopharyngeal swab induces no risk and no additional personal data beyond the usual information on 96 symptoms and duration required by the FOPH for all SARS-CoV-2 tests in Switzerland was collected 97 (Req-2020-00398). Due to the ethics waiver no informed consent had to be collected. 98 Sample collection 99 Test centers were advised to use their regular swab and virus transport medium (VTM)/universal 100 transport medium (UTM) for nasopharyngeal sampling. Transport media used by the centers included laboratory of the Institute of Medical Virology. 500 ul of NPS or saliva in VTM were diluted in 500 ul 118 of Nuclisens easyMAG Lysis Buffer (BioMérieux), centrifuged (2000 rpm, 5 min) and analyzed with 119 the Cobas SARS-CoV-2 IVD test (Roche) on a Cobas 6800. All testing for NPS and saliva was done 120 in parallel on the same day. SARS-CoV-2 detection was further quantified using SARS-CoV-2

136
Head-to-head comparison of saliva and nasopharyngeal swabs as material for SARS-CoV-2 detection 137 by  In our protocol we advised participants to collect approx. 0.5 ml saliva into a wide (30 ml, 30 mm 139 diameter) tube ( Figure 1). Initial attempts in a pilot experiment with smaller tubes (15 ml, 17 mm 140 diameter) showed that spitting into narrower tubes is problematic for some participants, leading to a 141 contamination of the outside of the tube with saliva in some cases. Sampling with the wider tubes was 142 in contrast unproblematic and thus deemed safe. Saliva sampling in children was found equally 143 unproblematic, children were collaborating and able to expectorate.

144
Our study included five different test sites to ensure that data are not skewed due to specific 145 procedures at one site. In Study Arm "Basic" (N = 835) saliva sampling was done with one-time throat 146 clearing followed by expectorating saliva one to two times. In Study Arm "Enhanced" (N = 352) 147 participants cleared their throat 3x times followed by spitting. Saliva was mixed with VTM 148 . CC-BY-NC 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 3, 2020. ; https://doi.org/10.1101/2020.12.01.20241778 doi: medRxiv preprint immediately after collection. The thus diluted material was unproblematic for further processing in the 149 laboratory, no complications in pipetting or invalid results due to the intrinsic viscosity of saliva or 150 congealing were observed.

151
High positive predictive agreement of SARS-CoV-2 detection in saliva and nasopharyngeal swabs 152 Adults and children that qualified for a regular SARS-CoV-2 test according to the FOPH and reported 153 to one of the participating test centers or emergency units were enrolled from October 20, 2020 to Median age was 35 with an age range of 5 -98 years. 89 participants were under the age of 18. The 156 majority of participants were symptomatic 71.9%. Median Days of symptoms ranged from 1 to 30 157 with a median of 2 days. The overall daily positivity rate of SARS-CoV-2 tests by RT-PCR during the 158 study period at our diagnostics unit ranged between 14% and 22%. The positivity rate amongst study 159 participants was 21%.  Table 2). To investigate if discordant results are due to inadequate 164 sampling, detection problems in the RT-PCR, or reflect true negatives in the respective sample 165 material, all discordant pairs were retested using an in-house RT-PCR for the E-gene in conjunction 166 with a GAPDH measurement to control for input. Mean levels for GAPDH input were Ct = 24.6 (SD 167 = 2.7) for NPS and Ct = 24.7 (SD = 2.1) for saliva. One false-negative saliva sample (E-gene Ct 19.7 168 in NPS) did not contain any material (GAPDH Ct > 40). Excluding this sample, the PPA in the NPS

170
Re-assessment with an in-house E-gene PCR confirmed all discordant results. For one case with a 171 negative NPS, a second swab was collected the following day. This sample showed a high viral load, 172 confirming an unsuccessful swab collection the day earlier.

173
Of note, in our head-to-head comparison both NPS (N = 1) and saliva (N = 5; N = 4 excluding the 174 sample that did not contain saliva) produced false-negative results in cases where the other specimen 175 showed a high viral load (Ct < 30) highlighting variability in collection for both specimens. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 3, 2020. ; https://doi.org/10.1101/2020.12.01.20241778 doi: medRxiv preprint self-evaluation ( Figure 4A). We observed a good positive percent agreement of saliva and NPS in 185 symptomatic individuals (PPA = 92.3%). In line with a trend to lower viral loads, i.e. higher Ct values 186 in absence of symptoms (asymptomatic median Ct 28.4; mild symptoms median Ct 23.7; strong 187 symptoms median Ct 21.6), the PPA was lower in asymptomatic participants (PPA = 84.2%). We  In the present study we sought to devise and evaluate a saliva sampling strategy that provides i) 202 representative sampling of virus containing material, ii) easy and safe sampling in adults and children, 203 iii) possibility for home collection, iv) straight forward processing in the laboratory. 204 We opted for a saliva collection procedure where participants clear their throat to first generate saliva 205 from the back of the throat and then expectorate the saliva into an empty container. We considered 206 clearing the throat important to sample material from the posterior oropharynx where SARS-CoV-2 207 sampling by oropharyngeal swabs is known to be efficient [34,35]. While gargling with saline or 208 buffer solutions has been suggested as a possibility to sample saliva from the deep throat [36, 37], we 209 rated this procedure as less operable as the gargling solution would need to be optimized for taste to be 210 accepted by individuals, could not include preservatives, and gargling itself may potentially generate 211 aerosols. In addition, gargling is not practicable for many smaller children for whom we in particular 212 sought to create increased possibilities for SARS-CoV-2 testing as NPS collection for children is often is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 3, 2020. ; https://doi.org/10.1101/2020.12.01.20241778 doi: medRxiv preprint marginal risk for transmission as suggested by contact tracing and in vitro culturing studies [38][39][40]. 219 Considering the observed PPA in detection, saliva may safely be envisaged as substitute for NPS 220 detection in a range of settings. Possible scenarios include i) sampling of children, ii) home collection 221 in quarantine, iii) test centers without trained medical personnel (e.g. schools, universities, 222 companies), iv) non-irritating alternative for persons that need frequent testing due to their occupation 223 or health status, v) fast large-scale screens in institutions (e.g. elderly homes). In situations where 224 besides SARS-CoV-2 other respiratory viruses, e.g., Influenza and RSV, need to be excluded, NPS 225 should, however, remain the standard material of choice as it allows rapid detection with multiplex-226 PCR from a single specimen. In addition, if SARS-CoV-2 infection has to be ruled out with highest 227 possible sensitivity (e.g. in transplantation), NPS should remain the standard procedure.

228
The majority of SARS-CoV-2 in saliva represents likely virus secreted from infected cells in the 229 nasopharynx and is not locally produced. Collecting material from the posterior oropharynx is thus 230 important. This is also highlighted in our study as the collection protocol with intensified throat 231 clearing shows a trend to increased PPA at low viral loads.

232
It remains possible that eating or drinking shortly before collection may decrease viral content in the 233 oral cavity and throat. In the present study, neither eating, drinking nor smoking was controlled as 234 study subjects came for an elective analysis by NPS and thus could only be informed about the saliva 235 sampling on site immediately before the collection. Abstaining from food and beverage uptake shortly 236 (1h) before saliva collection could be considered in forth-coming applications of saliva as test 237 material, as it may increase the efficacy of SARS-CoV-2 detection in saliva even further.

238
In summary, our analysis rates saliva as valid alternate specimen for SARS-CoV-2 detection by RT-239 PCR. Saliva collection is non-invasive, thus not strenuous for patients, does not need trained 240 personnel, allows collection at any location, and allows self-collection. Importantly, as we show here,  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 3, 2020. ; https://doi.org/10.1101/2020.12.01.20241778 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 3, 2020.    is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 3, 2020. ;https://doi.org/10.1101https://doi.org/10. /2020  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 3, 2020. ; https://doi.org/10.1101/2020.12.01.20241778 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 3, 2020. ; https://doi.org/10.1101/2020.12.01.20241778 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 3, 2020. ; https://doi.org/10.1101/2020.12.01.20241778 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 3, 2020. ; https://doi.org/10.1101/2020.12.01.20241778 doi: medRxiv preprint . CC-BY-NC 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted December 3, 2020. ; https://doi.org/10.1101/2020.12.01.20241778 doi: medRxiv preprint