Validation and testing of a method for detection of SARS-CoV-2 RNA in healthy human stool

Viral reference material

Heat inactivated SARS-CoV-2 (iCOV2) was used for spike-ins. A SARS-CoV-2 isolate obtained from BEI Resources (NIAID, NIH) was propagated in Vero E6 cells (ATCC, Cat# CRL-1586) cultured in Eagle’s minimal essential medium supplemented with 10% fetal bovine serum (Invitrogen) and 1% penicillin/streptomycin and L-glutamine. Harvested cell supernatants were inactivated in a water bath at 60°C for 1 hour. Inactivated virus stocks were diluted in sterile PBS (Gibco), aliquoted, and stored at −80°C. To minimize degradation from freeze-thaw activity, fresh freezer stocks were used for each experiment.

Copy number concentrations for the virus stocks were determined by droplet digital PCR (ddPCR) using the QX-200 system (Bio-Rad), the One-Step RT-ddPCR advanced kit for probes (Bio-Rad), and 2019- nCoV CDC ddPCR Triplex probe kit (Bio-Rad) following all manufacturer protocols. RNA was isolated from freshly thawed freezer stocks using the QIAamp Viral RNA mini kit (Qiagen) following manufacturer recommendations. Absolute quantification of viral genome targets was determined using Quantasoft software V. 1.7.4 (Bio-Rad).

Human fecal samples

Initial rRT-PCR protocol development, limit of detection experiments, and subsequent experiments were performed using human fecal material purchased from two vendors, Lee Biosolutions and OpenBiome. All fecal samples were collected by the vendors with informed consent from healthy donors between 18 and 50 years of age with no use of antibiotics for at least 30-days prior to sample donation. To ensure SARS-CoV-2 negativity, only samples collected and frozen by the vendors prior to January 2020 were purchased.

rRT-PCR assay conditions

The protocols for isolation of viral RNA from stool samples and RNA target detection were adapted from previously published work¹². Prior to RNA extraction, stool was diluted 1:5 (w:v) in sterile PBS, homogenized, and clarified by centrifugation at 4000 rcf for 20 minutes at 4°C. 140 μL of clarified supernatant was used as starting material for RNA extraction using the QIAamp viral RNA mini kit (Qiagen) spin-column protocol, with a final elution volume of 50 μL.

For the SARS-CoV-2 detection protocol, specimens were assayed for three unique targets in independent reactions using sequence-specific primers and fluorescent hydrolysis probes as described in the SARS-CoV-2 (2019-nCoV) CDC EUA protocol¹⁴. Two of the primer/probe sets, 2019_nCOV_N1 (N1) and 2019_nCOV_N2 (N2), target conserved regions within the nucleocapsid gene. A third set targeting the human RNase P gene (RP) was used as an internal amplification control. Primer/probe sets for each reaction were purchased together from IDT (2019-nCoV RUO kits).

Individual target amplifications were performed in 20 μL reactions consisting of 5 μL TaqPath™ 1-Step RT-qPCR Master Mix, CG (Life Technologies), 1·5 μL respective primer/probe mix, 8·5μL of nuclease-free water, and 5μL of extracted template RNA. The final concentrations of primers and probes were 500 nM of each primer and 125 nM of the probe. PCR reactions were performed according to the CDC EUA protocol with the exception of increasing the amplification denaturation step time from 3 to 15 seconds¹⁴. All PCR runs were performed on the CFX-Connect 96-well real-time PCR detection system (Bio-Rad) and real-time fluorescence data were analyzed using CFX Manager software v. 3.1 (Bio-Rad). For this protocol, amplification of either the N1 or N2 viral targets with a threshold cycle (Ct) below 40·0 was considered SARS-CoV-2 positive. Samples were classified as negative when both N1 and N2 failed to amplify with Ct < 40·0 and the RP gene target with Ct < 40·0 was successfully detected. Samples in which none of the targets amplified with Ct < 40·0 were considered invalid.

Analytical sensitivity

An approximate range for the expected limit of detection (LoD) was determined by testing stool specimens spiked with a 10-fold dilution series of iCOV2 material (data not shown). Subsequently, negative stool slurries from two independent donors were spiked with iCOV2 reference material from a 2- fold dilution series and tested in replicates of 20 (10 per donor stool), including independent RNA extraction for each replicate. At least one un-spiked stool from each donor was tested with each batch of extractions as a negative specimen control, along with no-template controls and positive controls for each reaction. The LoD was defined as the lowest concentration at which at least 95% of samples tested positive.

Leftover spiked stool material from these dilutions was stored at −80°C and later tested to investigate the effects of freezing SARS-CoV-2 positive samples at or near the LoD. Frozen samples at 1x and 2x the empirical LoD were thawed and tested in replicates of 8 (4 per donor) to determine the impact of freeze-thaw activity on low-copy-number specimens.

Performance evaluation

A total of 24 contrived positive specimens and 6 negative specimens were tested in duplicate rounds of testing. Positive specimens were contrived by spiking iCOV2 reference material into a negative stool matrix at approximately 1x, 2x, 5x, and 10x the LoD. Samples were prepared at the FDA and shipped over night on dry ice to Stanford University. All samples were analyzed in duplicate following the testing protocols described above. The recipient lab was blinded to the spike-in status of the specimens. Testing at Stanford was identical to the protocol used at FDA with the exception of the PCR system used (Applied Biosciences 7900 HT Fast Realtime PCR system).

Additionally, five stool specimens were obtained from four pediatric patients with recent SARS-CoV-2 positive respiratory tests (Table 3). RNA was extracted from stool at the Ann & Robert H. Lurie Children’s Hospital in Chicago, Illinois and shipped overnight on dry ice to the FDA for rRT-PCR analysis. All samples were tested in duplicate.

Viral RNA stability during sample storage

The stability of SARS-CoV-2 RNA was tested in stool in different storage buffers and at different storage temperatures. Four different storage media were tested including PBS, Cary-Blair media, Stool Transport and Recovery (STAR) buffer (Roche Diagnostics), and DNA/RNA Shield (Zymo Research). Stool slurries in each buffer (1·5 grams of stool to 5 mL of buffer) were spiked with approximately 100x the LoD of inactivated SARS-CoV-2. Aliquots of each slurry were stored at either −80°C, 4°C, or ambient temperature (∼21°C). RNA extractions were performed from each sample set immediately (day 0) and on days 1, 2, 3, and 7 of storage at 4°C and ambient temperature, and on day 7 only from samples stored at −80°C. Extracted RNA was stored at −80°C until the end of the experiment, when rRT-PCR was performed in duplicate using only the N1 target. This experiment was replicated using independently collected stool samples from the same donor, for a total of two biological replicates.

Data analysis

All reported average Ct values with accompanying standard deviations were calculated in Excel, and were inclusive of any detected Ct values, regardless of whether they met the 40 Ct cutoff for positive detection. Contrived clinical testing agreement rates and associated Clopper-Pearson confidence intervals were calculated using the Westgard QC online 2×2 contingency calculator tool following guidelines laid out in CLSI document EP12-A2¹⁵. Results of viral stability experiments were graphed and analyzed with mixed-effects models for repeated measures with multiple comparison testing in GraphPad Prism V. 8.4.0 (GraphPad Software). All reported p-values and confidence intervals are adjusted for multiple comparisons.

Analytical performance

To validate the limit of detection of this assay, we first spiked dilutions of iCOV2 into stool and performed replicate testing of our rRT-PCR assay. We found the LoD to be 600 viral RNA copy equivalents per mL of stool matrix (Table 1). After accounting for a 1:5 stool dilution used in this protocol, this translates to 3000 copies per gram of whole stool using this extraction protocol. At viral concentrations of 1200, 600, and 300 copies per mL of stool slurry, test positivity rates were found to be 95% (19/20), 100% (20/20), and 70% (14/20), respectively (Table 1). The human RP gene was used as an amplification control in all samples to ensure that the PCR reactions were working correctly. This target was detected in all 20 replicates tested from each dilution, along with all negative specimen control extractions tests. At 1x the LoD, N1 and N2 targets were positive in 19 and 16 of the 20 replicates, respectively (Table 1). Notably, N2 amplification above baseline fluorescence was detected in all 20 samples, however only 16 met the criteria for positive detection (Ct ≤ 40·0).

The LoD in frozen samples was not consistent with that of freshly spiked samples. Frozen samples at 1x the LoD had a test positivity rate of 75% (6/8) after one freeze-thaw cycle (Table 2), suggesting that freezer storage impacts SARS-CoV-2 RNA detection in stool with low viral copy numbers. However, frozen samples at 2x the LoD tested with 100% positivity (8/8), suggesting that the effect of freezing on LoD is relatively minimal.

Interlaboratory assessment

The reproducibility of this protocol between labs was tested using mock positive samples (iCOV2-spiked) and negative samples, which were frozen and shipped to the secondary site (Stanford University) for testing. The secondary site was blinded to the identity of these samples. Of the 30 contrived samples tested, only one was found to be invalid due to the non-amplification of all three targets. Of the 29 valid test results, 22 of 24 spiked samples tested positive, and 5 of 5 negative samples tested negative, for an overall testing agreement rate of 93·1% (CI 78·0 – 98·1). In a replicate round of testing for all samples, three of the contrived samples returned invalid results. Of the 27 valid replicate test results, 21 of 23 contrived positive samples tested positive and 4 of 4 negative samples tested negative, for an overall agreement rate of 92·6% (CI 76·6 – 97·9). All false-negative tests were in the concentration range of 1x to 2x the assay LoD, within which freezer storage can negatively impact SARS-CoV-2 detection (Table 2). Notably, there were no false-positive test results and none of the samples tested invalid or falsely negative in both rounds of replicate testing.

Effects of storage conditions Ct signal stability

The effects of storage buffer and temperature on viral RNA stability were assessed in four stool storage buffers by spiking stool suspensions in different storage buffers at 100x the LoD. We then performed rRT-PCR on the samples and compared N1 Ct values from RNA extracted at various timepoints from each storage condition (Figure 1a). At baseline (day 0), mean Ct values were significantly higher in spiked stools suspended in Cary-Blair medium (29·4) and STAR buffer (30·4) compared with both PBS (28·8) and DNA/RNA Shield (28·5), possibly reflecting differences in the efficiency of RNA recovery following resuspension in the different buffers. Comparing samples stored at −80°C for one week with their respective baseline values, Ct values increased significantly in both PBS (29·7) and Cary-Blair (30·8), while there was no effect of freezing in the protective STAR buffer and DNA/RNA Shield (Figure 1a).

• Download figure
• Open in new tab

Figure 1.

Stability of N1 Ct values in stool with different storage media and temperatures. A: Comparison of Ct values from freshly spiked stool (circles) with those of spiked stool frozen at −80°C for 1 week (triangles). Each symbol represents a single technical replicate, with biological replicates differentiated by solid and hollow symbol types. Bars represent simple mean and 95% confidence intervals. B: Ct values from stool stored at 4°C (solid) or ambient temperature (hollow). Symbols represent the mean of technical and biological replicates for a given time point and storage buffer, with 95% confidence intervals shown.

Note: Biological replicates were assayed on separate plates but analyzed in aggregate. Baseline (day 0) values in the top and bottom panels are the same data.

Among samples stored at 4°C for up to seven days, only those stored in DNA/RNA Shield exhibited Ct values increasing significantly above baseline, reaching significance on day 2 (p < 0·05) and a maximum Ct increase of 1·27 ± 0·29 over baseline on day 7 (Figure 1b; p < 0·001). There were non-significant increases in PBS and Cary-Blair Ct values at 4°C, each with a maximum increase at 7 days. Ct values for samples stored in STAR buffer decreased steadily compared to baseline, reaching a maximum difference of −0·65 ± 0·88 on day 7 (Figure 1b; p = 0·11). At ambient temperature storage, both PBS and STAR buffer had significant increases over baseline Ct values (Figure 1b). The maximum observed increase in STAR buffer was 1·92 ± 0·99 above baseline (p < 0·01) on day 7. In PBS, the maximum observed increase was 1·50 ± 0·04 Ct units above baseline at 7 days of storage (P ≤ 0·0001). Cary-Blair and DNA/RNA Shield samples each had non-significant increases in Ct values over baseline, each reaching maximum observed differences at 7 days (Figure 1b).

Assessment of assay performance on clinical specimens

To definitively show that this method could detect SARS-CoV-2 RNA in the stool of infected individuals, we obtained stool samples from individuals who had previously tested positive for SARS-CoV-2 (Table 3). Five stool samples from four pediatric patients with recent COVID-19 positive respiratory swab tests were tested using our assay. Three of the five stool specimens tested positive for SARS-CoV-2 RNA, representing two of the four patients.