An open source and convenient method for the wide-spread testing of COVID-19 using deep throat sputum samples

Informed consent

All study participants were enrolled and sampled in accordance to the University of Iowa IRB-01 approved protocol #: 202004568. All study participants gave consent, but we had a waiver of documentation of consent. Recruitment was conducted through advertisements, word of mouth, and referral from a physician caring for COVID-19 patients, with the patients agreeing for us to contact them by email or telephone. Clinical data and samples were only gathered after the study participant had acknowledged that they understood the study protocol and consented. All participant information and samples were gathered in association with study identifiers. After data analysis, we destroyed the link between the ID code and subject identifiers.

Collection of positive and negative samples from the community

The study was conducted prospectively. We collected samples from patients that were deemed positive or negative for SARS-CoV2. We aimed to collect a minimum of 30 positive samples and 30 negative samples in order to ensure that our 95% confidence interval for 30/30 would have a minimal of 89% for specificity and sensitivity.

For positive cases, we obtained samples from patients presenting with signs and symptoms of COVID-19, who underwent NP swab collection of specimens at the University of Iowa Hospitals and Clinics (UIHC). These samples were processed and tested using the CDC approved methods (EUA2000001). Subsequently, patients who tested positive at an FDA approved testing center were recruited and enrolled into our IRB-approved study after obtaining consent. Following enrollment, patients independently self-collected DTS samples. Samples were collected by the patient within 72 h of being tested positive at an FDA approved testing center. It should be noted that most samples were collected within 24 h of testing positive at an approved testing center. All samples were transported at ambient temperatures, and tested within 48 h of collection using the PSCS-SARS-CoV2 test. A positive in the PSCS-CoV2 assay was determined to be a cycle threshold < 40 for N1 and < 40 for a second SARS-CoV-2 marker, N2 or E. This threshold was set from earlier recommendations from CDC. An RP of < 34 was needed to ensure the integrity of the sample.

Negative cases were recruited based on the absence of symptoms including absence of a history of a fever for the past month and lack of contact with known positive patients. The subjects were provided instructions and independently self-collected DTS fluid. Nasopharyngeal swabs from the subjects were also collected by a trained medical provider. All samples were transported at ambient temperatures and tested within 48 h of collection using the PSCS-SARS-CoV2 test, unless noted in the results. A negative sample was determined to be a cycle threshold > 40 for N1 and > 40 for a second SARS-CoV-2 marker, N2.

Collection of field samples

Field samples were recruited from a local college campus. A positive was determined to be a cycle threshold < 40 for N1 and < 40 for a second SARS-CoV-2 marker N2 or E. Samples that did not reach positive criteria for both markers were scored as negative. Samples that were positive for one marker and negative for the other marker were scored as ambiguous and retested. Participants were notified of their result and asked to self-report their COVID results from an FDA approved testing site. For the samples used for the pooling strategy, individual samples were tested if they had a cycle threshold < 40 for N1 or < 40 for a second SARS-CoV-2 marker (N2 or E1). We used a different second SARS-CoV-2 marker, E1, in some cases to test the validity of our original N2 SARS-CoV-2 marker.

Patient self-collection of sample for SARS-CoV-2

PSCS-CoV2 is a deep throat sputum (DTS) fluid collection protocol and specimen kit, which allows patients to self-collect a sample without the physical assistance of or close contact with another individual, most importantly without the assistance of a healthcare worker. The fluid collected contains both sputum and saliva. The Two-Step Kit container allows for the inactivation of virus within the container once the patient has self-collected the sample. The containers were placed in a biohazard bag, and handled with caution. Samples were then processed in a certified laboratory in a BSL2+ hood to isolate the viral RNA for detection using real-time RT-PCR testing intended for the presumptive qualitative detection of nucleic acid from the SARS-CoV-2. We obtained samples from individuals diagnosed with COVID-19, or individuals negative for COVID-19, or individuals with signs and symptoms of infection who are suspected of COVID-19 or individuals exposed to COVID-19, such as healthcare workers, family members of COVID-19 patients, individuals living in care facilities, etc.

Materials

PSCS-CoV2 two-step protocol

1. Do not eat, drink, smoke, or chew gum for 30 min prior to use.

2. The subject was provided a Two-Step Kit, which contains instructions, one sealable biohazard plastic bag, one large sample collection tube, one VIS container, and a prelabeled sticker with the patient’s ID information to identify the sample.

3. The subject opens the kit and places the ID sticker on the side of the large sample collection container. The subject then removes the cap from the large collection container and sets the small VIS container to the side for later use in step 5.

4. The subject, while isolated from other individuals, snorts through the nose with mouth closed by taking a deep breath inward, then coughs to clear the throat attempting to bring up phlegm (phlegm was NOT a requirement for a good sample), and spits into the large collection container. Only a small volume was needed (1 spit was usually plenty, usually yielding ∼0.5 mL).

5. The subject then takes the VIS container, removes the lid, and pours the liquid into the large collection container. The subject throws away the small VIS container.

6. The subject caps the large collection container tightly and shakes vigorously for ∼30 s to mix the collected sample with VIS.

7. Subject places the container back in the provided sealable biohazard plastic bag. The sample was now ready for pick-up or drop-off.

Subsequent testing for SARS-CoV2 in self-collected samples obtained by the PSCS-CoV2 method: We use the classic single step phenol-chloroform RNA preparation protocol (Chomczynski & Sacchi, 2006), with some minor modifications, beginning with 500 µL of the collected sample (mixed with VIS). Once the RNA was isolated and was in dry pellet form, the pellet was resuspended in 20 µL of RNase-free water, 2.5 µL of the resuspended RNA was used for real-time RT-PCR. The remaining sample may be stored for replications, as needed.

Real-time RT-PCR

The specific nucleic acid sequences from the genome of SARS-CoV-2 [2019-nCoV_PH_nCOV_20_026 N gene for nucleocapsid phosphoprotein] are used for RT-PCR primers. Specifically, primers for real-time RT-PCR for N1, N2 and E sequences are used. These primers are provided by Integrated DNA Technologies (IDT) (Iowa City, IA, USA) in their 2019-nCoV CDC EUA Kit (IDT Catalog #10006606). This kit contains primers against two regions of the RNA sequence for the nucleocapsid phosphoprotein (referred to as N1 and N2), envelope protein (E) and the human RNase Protein (RP) as a positive collection control. We initially used a cDNA template control for the N1 and N2 sequences. However, early in the study, we were also able to acquire an RNA control containing the SARS-CoV-2 genome from a COVID-19 positive patient. We used this sample as a positive control. Sequencing of the RNA control confirmed that the sample was SARS-CoV-2.

The N1 and N2 are commonly used to test for SARS-CoV-2 with no concerns about alteration of analytical sensitivity characteristics of viral detection for the delta variant. The delta variant is not mutated in the N1, N2, or E primer or probe sequences.

Real-time RT-PCR was performed using the protocols as described in the CDC 2019-Novel Coronavirus (2019-nCov) Real-Time RT-PCR Diagnostic Panel (Catalog#2019-nCoVEUA-01; effective 3/30/2020) (Ref. 10), with the exception that a Bio-Rad CFX96 Real-Time PCR instrument was used, rather than the Applied Biosystems 7500 Real-time PCR instrument.

Viral inactivating solution assessment study

We used mouse hepatitis virus (MHV), strain A59, a beta coronavirus that is closely related to SARS-CoV-2. Fifty microliters of VIS, VIS-DTS (1:3), or PBS were mixed 1:1 with 50 µL MHV at varying concentration of viral titers and incubated for 10 min at ambient temperature. Live virus titers were determined using a plaque assay of HeLa cells expressing the MHV receptors as described in a previous paper (Grunewald et al., 2020).

Limit of detection study

A DTS sample was collected from a COVID-19 patient previously confirmed positive by an independent Clinical Laboratory Improvement Amendments (CLIA) approved lab. We isolated the RNA from this sample using the described phenol/chloroform extraction method. The RNA from this sample was diluted with DTS from a known negative subject at various dilutions to determine the limit of detection (LoD). The LoD is the greatest dilution of a DTS sample from a COVID-19 patient to still test as positive with our protocol. The various dilutions we tested were 1/10, 1/100, 1/10³, 1/10⁴ and 1/10⁵. The positive COVID-19 DTS sample was diluted in a known negative DTS sample. The RNA was extracted using the phenol/chloroform method and assayed using our real-time RT-PCR method (PSCS-SARS-CoV2). These samples were then assayed in triplicate, and we determined that 1/10⁴ was the LoD.

The 1/10⁴ dilution was then used to spike 20 aliquots of DTS at 2xLoD (1/5000 final dilution) from known negative subjects. These spiked samples were processed with the phenol/chloroform extraction method and assayed using the downstream assay (PSCS-SARS-CoV2) with primers and probes for N1, E and RP.

Data analysis

Data analysts and experimenters were blinded to the COVID status of the samples. The investigators collecting the sample were aware of the COVID status and unblinded the results after data analysis. Cycle threshold for determining the sensitivity and specificity of the samples were determined prospectively. The Wilson/Brown method was used to calculate the 95% confidence interval for sensitivity and specificity. The cycle threshold for the field data and pooling data were determined retrospectively, after the data were collected. Statistical analysis and graphing was performed using GraphPad Prism 8.2.0.

Clinical validation of SARS-CoV-2 using the patient self-collection of DTS method

To test the accuracy of the Patient Self-Collection of Sample-CoV2 (PSCS-CoV2), we recruited COVID-19 positive and negative patients. A total of 42 positives patients (confirmed positive from an independent CLIA approved laboratory) and 30 negative subjects (confirmed negative with nasopharyngeal swab samples) underwent sample collection using the PSCS-CoV2 collection kit as described previously in this paper. The DTS samples were tested using real-time RT-PCR as described in the methods. The experimenter and data analysts were blinded to the COVID status of the samples. There was a 100% agreement between the results (i.e., positive or negative) obtained from testing of self-collected samples and those obtained from nasopharyngeal swabs (Table 1). Hence, our method had 100% specificity (95% CI [89–100]) and 100% sensitivity (95% CI [92–100]). We show a more detailed assessment of the SARS-CoV2 positive and negative samples in Table S1.

We also serially obtained samples using the PSCS-CoV2 collection method from a COVID-19 positive individual. The patient remained positive for at least 10 days after the initial positive test (Fig. S2). Notably, the DTS samples show reduced copies of viral RNA (increasing cycle threshold) over time.

Limits of detection experiments using SARS-CoV2 RNA

The Limit of Detection (LoD) study established the lowest SARS-CoV2 viral concentration detectable on DTS collected using the PSCS-CoV2 method. LoD was determined in two phases. First, positive SARS-CoV2 RNA at concentrations of 1/10, 1/100, 1/10³, 1/10⁴ and 1/10⁵ were assayed in triplicate. Results of this experiment show that viral RNA for all dilutions were positive in triplicate (Table 2). Second, we selected the 1/10⁴ dilution as the preliminary LoD because the cycle threshold (Ct) was lower than 30 for all three markers, indicating higher stringency than 1/10⁵ dilution. The 1/10⁴ dilution was confirmed with 20 contrived replicates at final 1/5000 dilution (2xLoD), showing 100% detection of SARS-CoV2 as shown by Ct < 40 for N1 and E (Table 2).

Overall, the PSCS-CoV2 method has a limit of detection of at least a 1/10⁴ dilution of a DTS sample from a positive patient. We estimate that there are approximately 100 copies of the viral RNA in a 1/10⁴ dilution. This was based on comparison of a SARS-CoV2 viral RNA sample with a quantified viral cDNA (plasmid) sample using quantitative PCR. A plasmid positive control sample of 100 copies gave an approximate Ct of 25, similar to the Ct of the 1/10⁴ viral RNA dilution.

Assessment of virus inactivation solution

To ensure that the Viral Inactivating Solution does not affect the accuracy of our PSCS-CoV2 method, we performed an analytical equivalency study. We tested the samples collected in our specific self-collection container (containing virus inactivating reagents) versus those collected in conventional collection containers (without any virus inactivating regents). To test COVID-19 positive samples, we used 1/5000 of a DTS from a positive COVID-19 patient 92x LoD). A total of 10 replicates for COVID-19 positive samples and a total of 10 replicates for COVID-19 negative samples (Table 3). There was 100% agreement between the two containers (with and without VIS). 100% of the spiked samples were positive for SARS-CoV-2 (with and without VIS), and 100% of the non-spiked samples were negative (with and without VIS). Of note, the Ct values were not significantly different between No VIS and VIS samples (p > 0.05) for any of the markers (N1, E, and RP).

To confirm inactivation of the infectious virus with VIS, we used the plaque assay method using mouse hepatitis virus (MHV), strain A59, a beta coronavirus that is closely related to SARS-CoV-2. The plaque forming unit (PFU) is a measure of the live virus titer. Following our protocol methods, treatment with VIS alone reduced live virus titers by approximately 99.9% and with VIS/DTS by at least 99.9% (PBS control: 2.2 × 106 PFU/mL; VIS: 3.75 × 103 PFU/mL; VIS/DTS: < 1.25 × 103 PFU/mL) (Calculations are in the Raw Data file).

RNA stability using the PSCS-CoV2 method with viral inactivating solution

We performed RNA specimen stability experiments using spiked samples in Viral Inactivation Solution (VIS). Testing RNA stability was critical since DTS samples may be subject to extreme temperatures on transit to the diagnostic laboratory. So we incubated the DTS samples at various temperatures, and for a duration of time that samples could be in transit. We had three groups of samples, samples spiked with 10x LoD for SARS-CoV2 RNA, samples spiked with 2x LoD for SARS-CoV2 RNA, and samples negative for SARS-CoV2 samples. We tested the three groups at 40 °C for 56 h (Table 4), 25 °C for 56 h (Table 5), and 4 °C for 56 h (Table 6). There was a 100% agreements for all three groups in detecting N1, E, and RP.

Field data of the PSCS-CoV2 method from a college campus

We tested subjects from a college campus for SARS-CoV-2 during the Fall 2020 . A subject was deemed positive for SARS-CoV-2 if their sample had a cycle threshold < 40 for N1 and < 40 for a second SARS-CoV-2 marker, N2. Of the 306 subjects tested by our PSCS-CoV2 method, 10.4% tested positive.

To provide an approximation of the accuracy of our PSCS-CoV2 method, we collected the results from subjects who were subsequently retested at an independent CLIA approved laboratory. The subjects underwent a nasopharygeal swab within a week of their initial PSCS-CoV2 test. Release of the results to us was entirely voluntary by the subjects and 23 subjects volunteered (Table 7). Ten of twelve (10/12) samples with an N1 < 30 by our test were confirmed positive. One of eight (1/8) samples with an N1 > 30 were confirmed positive. In addition, three of three (3/3) samples that were tested negative, were confirmed negative.

We hypothesized that the majority of the samples with an N1>30 would retest as negative since the subjects would be at the tail end of the disease. With the virus titer already low, it would be even lower by the time the subjects were retested. The sample that was retested as positive with an N1 > 30 were likely due to the patients being at the beginning stages of their infection. This test was also subject to response bias since those likely to get retested were those who were noted as positive for COVID-19.

Pooling strategy

To increase the time and resource efficiency for COVID-19 testing, we employed a pooling strategy. First, we combined equal aliquots of samples into a new tube. Second, the pooled sample was then tested for SARS-CoV-2. Third, from the pools tested positive for SARS-CoV-2, indivdual samples were tested for SARS-CoV-2.

We mathematically developed a pooling strategy for pool size based on positivity rate. Let k be the pool size and p the SARS-CoV-2 positivity rate. With probability (1 − p)^k, the pool is negative. In this case one test was enough. Otherwise k +1 tests are needed (1 for the pool followed by k individual tests). The expected number of tests with pooling was (1 − p)^k+ [1 − (1 − p)^k](k + 1). Pooling was meaningful only when this number was smaller than k, the number of individual tests. This implies p < 1 − (1/k)^1/k. For pool size k=2,4,6, and 8, p needs to be less than 0.292, 0.292, 0.258, and 0.229, respectively. Interestingly, for k = 2 and 4, the thresholds are the same.

We also developed the following equation to describe the max number of tests one would have to perform in the worst case scenario. The assumption was that each positive sample ends up in its own pool. We based the equation on the same COVID-19 positivity rate (p), pool size (k) and total number of individual samples (s).

Max number of trials = (s/k) + (p*s*k).

For a more conservative approach in determining the pool size based on the p, we utilized this equation of “max number of trials”. To ensure that the “number of combined trials in the pooling method” was never more than the standard method of “testing every individual sample”, the equation would be the following: s < (s/k) + (p∗s∗k), or 1 < 1/k + (k∗p). The requirement for this equations is that s > 2k because there need to be enough samples to make multiple pools. In addition, the assumption is that $p<\frac{1}{k}$ since a large positivity rate would void the equation. For pool size k=2, 4, 6, and 8, p needs to be less than 0.250,0.188,0.139,0.109, respectively.

We explored pooling stategy of pools of four subjects (Pool A) and pools of two subjects (Pool B) noted in Table 8. Individual samples were tested in a pool that had a cycle threshold <40 for N1 or <40 for a second SARS-CoV-2 marker (E). In our on-field data of a college campus, we had 12 pools of A (pool size of four, average CT of RP was 20.9). Three pools of A were positive. Altogether, we identified four individual positive samples (8.3%). Therefore, instead of having to test 48 individual samples, we only had to test 24 samples (12 pools and 12 individual samples) to identify the four COVID-19 positive patients.

For Pool B, we had 14 pools (pool size of 2, average CT of RP was 20.7). Three Pools of B were were positive. Altogether, we identified four individual positive samples (13%). So instead of having to test 28 individual samples, we only had to test 20 individual sample (14 pool and six inidividual samples) to identify the four COVID-19 positive patients.