Saliva Pooling Strategy for the Large-Scale Detection of SARS-CoV-2, Through Working-Groups Testing of Asymptomatic Subjects for Potential Applications in Different Workplaces

Objective:

To perform an improved large-scale SARS-CoV-2 detection on pooled tests of asymptomatic workers.

Methods:

qRT-PCR validation of the SARS-CoV-2 detection in salivae samples and saliva pools and working-group saliva pooling and testing for SARS-CoV-2.

Results:

We found a high Cycle threshold correlation (r = 0.9099) between swabs and saliva samples. Then, through the pooling strategy, we detected that 18/360 (5%) of individual saliva samples were SARS-CoV-2 positive. Saliva-pooling efficiency (360 of test sample/30 individual PCR) was higher (5.45) than the reported for swabbing group-testing and we spared 82% of the PCR reagents as well as sampling and personal protection equipment.

Conclusion:

Through this simplified and less expensive procedure, we detected in a short time asymptomatic-infected SARS-CoV-2-carriers that were isolated from their co-workers, thus, this methodology can be implemented in different workplaces to ensure consumers that employees are not infectious.

Learning Objectives

• Discuss the potential advantages and challenges of large-scale pooled saliva testing for SARS-CoV-2 in workplace settings.

• Summarize the new findings on detection of SARS-CoV-2 in saliva samples, compared to nasopharyngeal and oropharyngeal swabs.

• Discuss and summarize the findings of the authors’ pooled saliva testing strategy for workplace detection of SARS-CoV-2.

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the responsible of the human coronavirus disease 2019 (COVID-19) and the associated 2020 pandemic, that began at Guangdong, China.¹ Phylogenetic analyses of this virus proved that it has little similarity (80%) to other previously characterized SARS-CoV viruses, therefore the clinical and epidemiological characteristics of this pathogen are not fully known to this day.²

As of February 1, 2021, the infection of SARS-CoV-2 caused 103,369,000 laboratory-confirmed COVID-19 cases with more than 2,222,000 related deaths have been reported worldwide, being the USA, Brazil, India, and Mexico the most affected countries. In the latter one, until February 1, 2021, there are more than 1,860,000 confirmed COVID-19 cases including over 159,000 related deaths.³ Despite these alarming numbers, the COVID-19 true prevalence rate remains unknown, which is an important issue to consider, because its detection is the basis for SARS-CoV-2 infection prevention and control, through the isolation of infected patients.⁴ Considering these issues, it is relevant to develop new methods and strategies to detect SARS-CoV-2 in a simplified, fast, trustworthy, and large-scale method.

The most common clinical manifestations of COVID-19 are not specific, neither pathognomonic thus many are shared with other respiratory diseases. Among these symptoms are fever, dry cough, fatigue, and few patients may develop a rhinorrhea, sore throat, and non-respiratory manifestations such as arthralgia, myalgia, diarrhea, and dyspnea.⁵ Together with the lack of symptomatology specificity, not all the patients develop these symptoms, and some cases only develop milder symptoms such as low fever and slight fatigue.⁴ Hence, some of the infected subjects could be unaware that they are infected but may still carry and spread the virus. For this reason, a sensitive and specific method of SARS-CoV-2 detection is indispensable, and the most reliable method for the detection of active viral infection is through reverse transcription of the extracted RNA and quantitative PCR (qRT-PCR).^6,7

To perform the aforementioned qRT-PCR for SARS-CoV-2 detection, the international³ and national (Institute for Epidemiological Diagnosis and Reference -Instituto de Diagnóstico y Referencia Epidemiológicos “Dr Manuel Martínez Báez”, InDRE, Mexico) health authorities established the use of nasopharyngeal and/or oropharyngeal swabbing, as a reliable source of human and viral RNA. However, sample collection requires trained personnel,⁸ as well as sophisticated personal protection equipment (PPE) to avoid the contagion risk,⁴ and a special Viral Transport Medium (VTM). All these problems together with the shortage in laboratory equipment, have made the COVID-19 diagnosis more complicated and expensive.⁹ In this sense, several research groups have proposed the use of saliva samples as an alternative to the nasopharyngeal and/or oropharyngeal swabbing, since it is a non-invasive collection method, it does not require highly trained staff, and is not necessary the use of VTM, moreover, it minimizes the risk of infection to the personnel handling the samples and healthcare workers.^9-11 Aside from these advantages, saliva samples have other benefits such as that it is easily accessible, transportable, and storable. For all these reasons, it is an excellent candidate for large-scale sampling in COVID-19 in asymptomatic subjects.^9-12

In the same way that the efforts have been encouraged to improve the SARS-CoV-2 detection, new challenges have been raised. Every day COVID-19 diagnosis implies a single sample process that limits the number of processed samples. On the other hand, the detection of asymptomatic population that plays an important role in SARS-CoV-2 spreading, is essential for the control and prevention of this disease, as well as the screening of healthcare workers¹³ to ensure safe work environments in different areas of labor, and has been reported that the pooling methods rise the throughput, cut costs while keeping the sensibility of the SARS-CoV-2 detection test.¹³ The pooling strategy or group testing permits the identification of carriers in an N population with a lower number of tests than N.¹⁴ In addition, the use of pools would allow faster sampling of different work areas of workplaces (hotels, schools, laboratory), and to detect these subjects quickly and safely. On the other hand, the use of individual tests raises the detection time in different work areas, which would be slow, causing a greater spread of the infection among the same workers and the general population. In this sense, one of the most used and simplest schemes is the Dorfman pooling, in which all samples are assigned to a batch, these pools contain the same number of samples and if one of these is positive, all the samples within this pool are individually re-tested.¹⁴

Therefore, the first aim of this study was to compare and validate the sensibility of detection of SARS-CoV-2 in saliva samples in comparison to nasopharyngeal and oropharyngeal swabbing, through qRT-PCR. The second objective was to implement a strategy to detect the presence of SARS-CoV-2 at a large-scale in asymptomatic subjects.

MATERIALS AND METHODS

Sample Collection

All study participants signed the informed consent before samples, clinical data, and demographic information, were collected. This study was approved by the local bioethics commission registry number (CEBN/03/20).

For swabs and saliva samples collection, subjects were queried to avoid eating food, drink water, and teeth brushing, at least 4 hours before sampling.

Swabbing

For swab sampling, a flexible swab was passed through each subject's nostril reaching the nasopharynx. Another flexible swab was introduced through the mouth to reach the oropharynx. Both swabs were placed in the mucosa while gently circling for some seconds, then removed while rotating and placed in 2.5 mL of the sterile VTM. Sterile VTM (pH 7.10) was prepared with Hank's balanced salt solution (HBSS) (ThermoFisher Scientific, Cat n° 14190) supplemented with gentamicin sulfate (4 mg/mL) (ThermoFisher Scientific, Cat n° 15750078 penicillin/streptomycin (50,000 U/50,000 μg/mL) (ThermoFisher Scientific) (Cat n° 15140148), amphotericin B (0.4 mg/mL) (ThermoFisher Scientific) (Cat n° 15290018) and bovine serum albumin (5%) (ThermoFisher Scientific) (Cat n° 15561020).

Saliva

Saliva samples were collected from the subjects by asking them to repeatedly spit into a sterile 120 mL urine collection cup until a minimum of 2 mL was collected. Then were stored at −20°C overnight and processed for RNA viral isolation the day after they were collected.

qRT-PCR Procedure

Methods were validated by the Mexican health authorities (Mexican Health Ministry —Secretaría de Salud de México—, and InDRE—). All reagents, kits, and procedures were approved by these authorities. Inactivation and total RNA extraction were performed with RNA extraction by QIAmp Viral RNA Mini Kit (Qiagen, Cat No./ID: 1020953 USA, Germantown) with 140 mL of VTM from swabbing or saliva. qRT-PCR procedure was performed according to the Berlin protocol with modifications.¹⁵ Briefly, one-step qRT-PCR was performed with StarQ One-Step qRT-PCR (Qiagen, Cat No./ID: 210210, USA, Germantown kit, with the extracted RNA from both types of samples. And for the SARS-CoV-2 detection for viral E gene E_Sarbeco_Forward: ACAGGTACGTTAATAGTTAATAGCGT, E_Sarbeco_Reverse: ATATTGCAGCAGTACGCACACA, TaqMan probe E_Sarbeco_P1: FAMCACTAGCCATCCTTACTGCGCTTCG-BBQ), and as a control RNase P gene (RNAseP Forward: AGATTTGGACCTGCGAGCG, RNAseP Reverse, GAGCGGCTGTCTCCACAAGT, TaqMan probe RNAseP P1, FAM-TTCTGACCTGAAGGCTCTGCGCG-BHQ1).^15,16 qRT-PCR was performed with 5 μL (≈70 ng/μL) of extracted RNA in a total 25 μL reaction. All samples were analyzed with ABI Prism 7500 Sequence Detector System (Applied Biosystems) with the following protocol: 50°C for 15 minutes, 95°C for 2 minutes and then 45 cycles of 95°C for 15 seconds 82°C and 60°C for 30 seconds.

In all cases human gene (RNAseP) amplification was used as an internal control, and samples were considered as positive if the number of cycles needed for the fluorescent signal to cross the threshold cycle known as Cycle threshold (Ct) value was equal or lower than 38.

Intra-Laboratory Validation

To assess the efficacy of SARS-CoV-2 detection through PCR, we prospectively compared between paired nasopharyngeal/oropharyngeal swabs and saliva samples from 27 subjects, all SARS-CoV-2 positive.

The first step to validate the quality of the samples was achieving the RNAse P gene amplification, thereafter, the amplification of E gene in SARS-CoV-2 positive samples.

Pooling Approaches (Large-Scale Strategy Validation)

The first pooling strategy was to mix different proportions of saliva from subjects negative or positive for SARS-CoV-2 E gene through nasopharyngeal/oropharyngeal swabbing, and then they were compared. After positive/negative swabs confirmation, saliva samples were mixed as follows: the negative pool or Pool A (10 negative samples); positive pools or Pool B (1 positive/9 negative salivae) and Pool C (9 positive/1 negative saliva). The final volume of each saliva pool was 140 μL, this means that each one contains 14 μL of 10 different saliva samples. This final volume was processed as above mentioned for RNA viral extraction and PCR procedures. Single swabs and saliva from SARS-CoV-2 positive subjects were used as positive controls, to compare the sensibility of this procedure.

In all cases, the amplification of the human gene (RNAse P) was used as internal control. A pool was considered as positive if the SARS-CoV-2 E gene was amplified (viral gene).

Implementing Large-Scale Pooling Strategy

The first issue to consider in this methodology was the detection and elimination of evident SARS-CoV-2 infected subjects, thus increasing the efficiency of the test. For this purpose, we implemented a test with exclusion criteria discarding anyone with the following symptoms: cough, fever, dyspnea, headache, rhinorrhea, myalgia, arthralgia, odynophagia, chills, chest pain, anosmia, dyspepsia, and conjunctivitis. After symptomatic individual exclusion, at least 2 mL of saliva from 360 asymptomatic subjects were collected. The second issue to improve the test efficiency was the pools assembling, in which all the individuals sharing the same workplace were included in the same pool. This is important since it should be expected that virus transmission takes place in more intimate cohabiting interactions (meeting work, intra-family or school) thus one infected subject could spread the disease to the people with who interact closely, such as is usually happen with co-workers.

Once samples were obtained, they were stored at 4°C until processing. From these samples, 36 pools with a final volume of 140 μL were mixed. Each pool contained 14 μL of 10 saliva samples that were processed as aforementioned. A detailed workflow scheme is shown (Fig. 1)

FIGURE 1.

Representative scheme of saliva pools for detection of E Gene from SARS-CoV-2.

Internal control was also considered with pool amplification of RNAse P gene.

A negative result was interpreted as that all samples in the pool were negative, while a positive result implies that at least one sample in the pool was positive. If the pool viral gene was amplified with a Ct below 38, individual samples within the pool were defrosted and tested individually.

Test Efficiency

The efficiency is the expected number of samples evaluated through a single RT-PCR reaction, it is affected by the pool size and the disease prevalence rate, and can be calculated as empirical efficiency through the following formula¹⁴:

$$\text{Empirical}\hspace{0.17em}\text{efficincy}=\frac{\text{Total}\hspace{0.17em}\text{number}\hspace{0.17em}\text{of}\hspace{0.17em}\text{tested}\hspace{0.17em}\text{samples}.}{\text{Total}\hspace{0.17em}\text{Number}\hspace{0.17em}\text{of}\hspace{0.17em}\text{actual}\hspace{0.17em}\text{RT-PCR}\hspace{0.17em}\hspace{0.17em}\text{reactions}\hspace{0.17em}\text{performed}.}$$

Statistical Analysis

Data were charted as mean ± SD. To compare the differences between two groups the Student t test was used and t test with Welch's correction was applied when needed (groups with different variances determined through an F test). Meanwhile, to compare the differences between a non-parametrical Kruskall–Wallis test with multiple comparisons were used. The differences were considered significant when P ≤ 0.05. All the analyses were calculated with Prism 6^® software (GraphPad Sofware Inc.).

RESULTS

The nasopharyngeal/oropharyngeal swabs and the saliva are a reliable sources of viral and human RNA.

The overall positivity between the matched swabs and saliva (different samples) was 100% (27/27) of the virus detection (Fig. 2A), this means that all paired swab/saliva samples were positive for SARS-CoV-2 E gene. No statistical differences between the average Ct value 25.48 ± 5.3 (17 to 34) for swab and 27.59 ± 5.98 (18 to 37) for saliva were found. Correlation of viral load had a high positive correlation among samples (r = 0.9099) (Fig. 2B). Finally, the average difference amid swab and saliva Ct value was 2.55. Of these swabs versus saliva Ct values 3 remained with no change and in 5 saliva samples Ct was lower than its respective value from the swabs (Fig. 2C).

FIGURE 2.

Detection of SARS-CoV-2 in paired nasopharyngeal/oropharyngeal swabs vs saliva samples. (A) Comparison between SARS-CoV-2 E gene Ct values from swab (red circles) and saliva (blue squares) samples. Red (swabs) and blue (saliva) lines represent the mean ± SD of the data from 27 paired samples. Statistical significance was calculated using the t test. (B) Pearson correlation coefficient test between swab and saliva samples (r = 0.9099). (C) Graphic representation of individual viral Ct values of paired swab/saliva samples obtained from the same subject. n = 27. Ct, threshold cycle. Ct = 38 is the cut-off for a positive result.

Saliva can be used in a large-scale method for SARS-CoV-2 detection.

Thereafter, it was determined if saliva could be used as a sample for a large-scale pooling procedure, through the mixing of saliva from negative and positive subjects to SARS-CoV-2 E gene from nasopharyngeal/oropharyngeal swabbing. Thus, positive/negative confirmed swabs samples were mixed, and as expected, the negative pool (Pool A—10 negative salivae) presented a mean Ct above the cut off value (43.33 ± 2.51). Meanwhile, in the pools with positive saliva samples, SARS-CoV-2 E gene was detected; remarkably, it was observed that with one positive saliva sample the Ct value of Pool B was higher (26.67 ± 1.52) than the pool with nine positive saliva samples (22 ± 1.7), which were only 4.67 cycles, and no statistical differences were observed (Fig. 3). No differences were found among individual swabs/saliva positive controls and the Pools B and C.

FIGURE 3.

Validation of pooling strategy. Viral Ct of SARS-CoV-2 E gene from individual nasopharyngeal/oropharyngeal swabs (control single samples) and saliva pools. Pool A: 10 negative saliva; Pool B: 1 positive/9 negatives; Pool C: 9 positive/1 negative. The red color indicates the nasopharyngeal swabs and the blue color represents the saliva sample of the patients, the Ct was determined by qRT-PCR. The dotted line shows the cut point (Ct < 38) SARS-CoV-2 negative Graph represents the mean ± SD of 3 data. Statistical significance was calculated using a t test. Ct, threshold cycle. Ct = 38 is the cut-off for a positive result.

Thus, the present pooling strategy with saliva samples is a trustworthy method for SARS-CoV-2 detection.

Additionally, it was implemented a protocol for pooling 360 saliva samples from asymptomatic subjects into 36 pools. In all pools, RNAse P gene was amplified, and only in 3/36 (8.3%) the SARS-CoV-2 E gene was amplified, from these the positive pools were the number 3, 4 and 31 (Fig. 4).

FIGURE 4.

Implementation of pooling strategy. Viral Ct of SARS-CoV-2 E gen from 36 pools within 10 salivas from asymptomatic subjects. Graph represents the Ct value of each pool (1–36). n = 36. Ct, threshold cycle. Ct = 38 is the cut-off for a positive result.

After determining the overall pool positivity, we defrosted and tested the 30 saliva samples corresponding to the 3 positive pools (3, 4, and 31). In all individual defrosted salivae, we were able to amplify the RNAse P gene, thus, these samples were viable. Additionally, the SARS-CoV-2 E gene was amplified in 18/30 (60%) of individual saliva tested (Fig. 5). The average Ct values of the positive samples within the 3, 4, and 31 batches, were 26 ± 3.9, 25.85 ± 4.8, and 23 ± 1.4, respectively; the above suggests that there is a difference between the Ct of the pool and the individual average of 4, 1.85, and 5 (Table 1).

FIGURE 5.

Viral Ct detected in saliva obtained from SARS-CoV-2 positive pools. Ct of E gene in individual saliva samples of 3, 4 and 31 positive pools. Graph represents the Ct value of each subject. n = 30. Ct, threshold cycle. Ct = 38 is the cut-off for a positive result.

TABLE 1.

Cycle Threshold (Ct) Value Differences Among Positive Individual Salivae Samples

Batch	No. of Positive Saliva/Pool	Ct of Positive Saliva Pool	Ct of Positive Individual Samples	Ct Difference	Average Ct Difference
3	7/10	22	26 ± 3.9	4	3.6
4	7/10	24	25.8 ± 4.8	1.8
31	4/10	28	23 ± 1.4	5

Test efficiency was improved with the 10-saliva samples pooling strategy.

The empirical efficiency of the present pooling method was calculated through the Dorfman formula (Dorfman, 1943) and 360 saliva samples were considered as the total number of tested samples, 36 (pools), and 30 (individual) RT-PCR reactions. Thus, we obtained 5.45 as the empirical efficiency of the test. This is a statistical approach that allows us to eliminate unwanted interference or undesired errors and to obtain accurate results.

DISCUSSION

Nowadays, the rapid and opportune SARS-CoV-2 detection allows stopping the viral contagion chain, which is necessary to control and prevent the COVID-19. Saliva is a sample that can be used in a large-scale methodology because the process for its collection is fast than the one for the swabbing procedure, keep personnel contagion risk at a minimum, is easy to obtain, manage and store, and avoid the need for additional sampling material and medical training.¹⁷ Another advantage of the saliva sample for qRT-PCR viral detection is that we avoid the use of VTM, which is important because we spare the cost of this media and improve the perform of the test, because some of the components in this VTM, is reported that could inhibit the PCR reaction.¹⁸

The first issue to consider when changing the source of RNA is the sensibility of the test. Different studies have demonstrated that saliva is more sensitive for SARS-CoV-2 molecular detection, in comparison with the samples obtained from VTM nasopharyngeal or oropharyngeal swabs.^9-11 In this sense, the overall SARS-CoV-2 E gene positivity detected in swabs were the same as those found in saliva samples (27/27) (Fig. 2A), with a high positive correlation (r = 0.9099) (Fig. 2B). Despite the above correlation, the Ct value was higher in saliva (27.59 ± 5.9) as compared to swabs (25.48 ± 5.3), this is probably associated with the defrosting process.¹⁹ Nevertheless, such difference was acceptable since in all matched samples SARS-CoV-2 E gene was successfully amplified. As mentioned above, saliva is reported as a better sampling test for COVID-19 diagnostic and is associated with a lower Ct values/higher viral loads obtained, but it is necessary to consider some particularities in these studies. As an example, Iwasaki et al¹⁰ and Wyllie et al¹¹ tested for SARS-CoV-2 detection using only one swab of nasopharyngeal origin, which might reduce the sensitivity of the test; meanwhile, Iwasaki et al used saline solution instead VTM.^10,11 In both cases, the reduction of swabbing quality could reduce the difference between swabs and saliva samples with respect to viral loads. Thus, the obtained Ct value of 2.55, constitutes an acceptable difference if we consider that nasopharyngeal/oropharyngeal swabbing is an invasive and painful procedure with an increased risk to sampling personnel that requires special personnel training and more expensive due to safety equipment requirement, the swabs and VTM¹¹ for each patient.

Once the saliva test proved to be a reliable source of SARS-CoV-2 RNA, it was performed a pooling method by mixing different salivae proportions from positive and negative subjects confirmed through SARS-CoV-2 in nasopharyngeal/oropharyngeal swabs. The results showed that a positive sample within a batch can be detected with the minimal quantity of 14 μL of saliva, in a mixture with another nine negative salivae (final volume of 140 μL). This in concordance with some previous authors who reported that through qRT-PCR can be detected a copy of viral RNA in 1 μL of sample, being crucial for the detection of SARS-CoV-2 in the beginning of the onset symptoms (13 to 20 days later).^20,21 On the other hand, the pools with negative samples were found negative, as well as the positive pool were found positive. These results would allow health care workers performing a group-testing scheme with saliva as a reliable human and viral RNA source.

In addition, a large-scale pooling strategy was implemented with 360 samples mixed into 36 pools, from which 3/36 (8.3%) were found positive with 22, 24, and 28 Ct values (Fig. 4; Table 1). When individual analysis of the 30 samples of the 3 positive pools was performed, we identified the 7, 7, and 4, salivae with SARS-CoV-2 viral RNA, respectively. In the present pooling scheme, we were able to detect individual samples with Ct values above 35, which indicates that low viral loads can be determined through this procedure. Likewise, the high Ct values detection discards false negatives, since it is possible to determine low concentrations of SARS-CoV-2, thus being a sensitive test.

Through the present large-scale pooling scheme it was possible to analyze 360 samples using the resources equivalent to perform 36 regular swab tests, thus pooling implementation in asymptomatic subjects proved to be a successful methodology for large-scale detection of asymptomatic positive-spreading individuals, despite of the difference in Ct values observed between pools and individual saliva samples of 3.6 Ct. This was in concordance with the accepted loss of 3 Ct considered as a minor and clinically acceptable interchange with the disadvantages associated to the shortage of commercial nasopharyngeal/oropharyngeal swabs availability and cost, the accessibility and easy storing that saliva samples could bypass this problem.²²

Dorfman R test empirical efficiency is a crucial issue to ensure viability of the method, in this sense, the present methodology eliminates imprecisions and produces statistically accurate values. The detection of defective members in large populations¹⁴ regarding SARS-CoV-2 pooling scheme, can be explained as the tests are spared in the present method. In this 10-salivae pooling scenario that was performed, the empirical efficiency was 5.45, a higher value than the one reported previously for 4.587 and 2.377 for the 8- and 5-swabs pools, respectively. This might be associated to the symptomatic subject exclusion and with the prevalence rate, which in asymptomatic population is lower than 1%, being the pooling SARS-CoV-2 test efficiency among 5 and 7.5 a value that matches with the one we found (5.45).¹³

Overall, we use 66 of the 360 reactions that usually are used for SARS-CoV-2 detection, thus we spare the 82% of the PCR reactions, as well as the column and reagents for inactivation and extraction, the VTM and the swabs used for nasopharyngeal/oropharyngeal swabbing.

Our results showed that saliva is a reliable biological sample for detecting SARS-CoV-2 viral RNA and it can be used as a useful tool for massive diagnostic method, since 360 individuals can be analyzed in a few hours, obtaining reliable results that allowed the making preventive decisions, such as the isolation of asymptomatic subjects positive to SARS-CoV-2.

This large-scale protocol could be a useful methodology for testing in hospitals, schools, institutions, and companies, to identify the SARS-CoV-2 asymptomatic carriers that could potentially spread COVID-19. Moreover, the results of the present large-scale test could be accessible in a short time that can be one or two days from sample collection.

CONCLUSIONS

In conclusion, saliva pooling methodology can be useful to perform a group-testing large-scale detection of SARS-CoV-2 in health workers as well as a reliable test for workplaces. This method is useful to identify, isolate and prevent the spread of SARS-CoV-2 in the population.