US CDC Real-Time Reverse Transcription PCR Panel for Detection of Severe Acute Respiratory Syndrome Coronavirus 2

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was identified as the etiologic agent associated with coronavirus disease, which emerged in late 2019. In response, we developed a diagnostic panel consisting of 3 real-time reverse transcription PCR assays targeting the nucleocapsid gene and evaluated use of these assays for detecting SARS-CoV-2 infection. All assays demonstrated a linear dynamic range of 8 orders of magnitude and an analytical limit of detection of 5 copies/reaction of quantified RNA transcripts and 1 x 10−1.5 50% tissue culture infectious dose/mL of cell-cultured SARS-CoV-2. All assays performed comparably with nasopharyngeal and oropharyngeal secretions, serum, and fecal specimens spiked with cultured virus. We obtained no false-positive amplifications with other human coronaviruses or common respiratory pathogens. Results from all 3 assays were highly correlated during clinical specimen testing. On February 4, 2020, the Food and Drug Administration issued an Emergency Use Authorization to enable emergency use of this panel.

SARS-CoV-2 was isolated in Vero cells from a respiratory specimen from an early US COVID-19 patient (8). We measured infectious virus titer of virus stock by 50% tissue culture infectious dose (TCID 50 ) (2.01 × 10 6 TCID 50 /mL) and inactivated the virus by gamma irradiation. The inactivated stock virus was used as reference material for assay evaluation. Cell culture stocks or clinical specimens containing other respiratory viruses/bacteria were used to evaluate assay specificity. Ten nasopharyngeal swab samples that had been collected in 2018 (before COVID-19) and were negative for respiratory viruses were also available for assay specificity evaluation.

Specimen Processing and Nucleic Acid Extraction
We extracted total nucleic acid from 120 μL of respiratory specimens by using the EZ1 DSP Virus Kit (QIAGEN, https://www.qiagen.com) following the manufacturer's instructions and collected 120 μL elution volumes. We processed sputum specimens by adding equal volumes of 10 mM of freshly prepared No-Weigh Thermo Scientific Pierce dithiothreitol (Thermo Fisher Scientific, https://www.thermofisher.com) and incubating them at room temperature for 30 min with intermittent mixing or until the specimens were sufficiently liquefied for extraction. We processed stool specimens by preparing 10% suspensions by adding 100 μL of liquid stool or ≈100 mg of solid stool to 900 μL of phosphate-buffered saline, pH 7.4 (Thermo Fisher Scientific), pulse vortexing for 30 s, and centrifuging at 4,000 × g for 10 min at 4°C. We then carefully removed the clarified supernatant for extraction. To demonstrate successful nucleic acid recovery and reagent integrity, we extracted human specimen control consisting of cultured A549 cells concurrently with the test specimens as a procedural control. We either tested extracts immediately or stored them at <-70°C until use.

Primers and Probes
We aligned the N gene sequence from the publicly available SARS-CoV-2 genome (GenBank accession no. MN908947) with other coronavirus sequences available from GenBank by using MAFFT version 7.450 implemented in Geneious Prime (Geneious Biologics, https://www.geneious.com). We designed multiple primer/probe sets targeting regions in the 5′, middle, and 3′ regions of the N gene sequence with the aid of Primer Express software version 3.0.1 (Thermo Fisher Scientific). We selected 3 candidate gene regions, designated N1, N2, and N3, for further study (Table 1). N1 and N2 were designed to specifically detect SARS-CoV-2, and N3 was designed to universally detect all currently recognized clade 2 and 3 viruses within the subgenus Sarbecovirus (4), including SARS-CoV-2, SARS-CoV, and bat-and civet-SARS-like CoVs. BLASTn (http://www.ncbi. nlm.nih.gov/blast/Blast.cgi) analysis demonstrated no major combined similarity of primers and probes of each assay with other coronaviruses (OC43, 229E, HKU1, NL63, and Middle East respiratory syndrome coronavirus [MERS-CoV]) or microflora of humans that would potentially yield false-positive results. We synthesized all primers and probes by using standard phosphoramidite chemistry techniques at the CDC Biotechnology Core Facility. We labeled hydrolysis probes at the 5′ end with 6-carboxy-fluorescein (FAM) and at the 3′ end with Black Hole Quencher 1 (Biosearch Technologies, https://www.biosearchtech.com).

In Vitro RNA Transcript and Viral Template Control
Double-stranded DNA containing a 5′-T7 RNA polymerase promoter sequence (TAATACGACTCAC-TATAGGG) and the SARS-CoV-2 complete N gene sequence was synthesized by Integrated DNA Technologies (https://www.idtdna.com). We transcribed the DNA by using the MEGAscript T7 Transcription Kit (Thermo Fisher Scientific). The RNA transcripts were purified by using the MEGAclear Transcription Clean-Up Kit (Thermo Fisher Scientific) and quantified with a Qubit fluorometer by using a Qubit RNA HS Assay Kit (Thermo Fisher Scientific). We used the RNA transcript for subsequent assay evaluation and positive template control.

rRT-PCR Assay
We performed all rRT-PCR testing by using the Taq-Path 1-Step RT-qPCR Master Mix, CG (Thermo Fisher Scientific). Each 20-μL reaction contained 5 μL of 4X Master Mix (Thermo Fisher Scientific), 0.5 μL of 5 μmol/L probe, 0.5 μL each of 20 μmol/L forward and reverse primers, 8.5 μL of nuclease-free water, and 5 μL of nucleic acid extract. We conducted amplification in 96-well plates on an Applied Biosystems 7500 Fast Dx Real-Time PCR Instrument (Thermo Fisher Scientific). Thermocycling conditions consisted of 2 min at 25°C for uracil-DNA glycosylase incubation, 15 min at 50°C for reverse transcription, 2 min at 95°C for activation of the Taq enzyme, and 45 cycles of 3 s at 95°C and 30 s at 55°C. We set the threshold in the middle of exponential amplification phase in log view. A positive test result was defined as an exponential fluorescent curve that crossed the threshold within 40 cycles (cycle threshold [C t ] <40).

Test Algorithm
For routine specimen testing, we ran all 3 SARS-CoV-2 assays simultaneously along with the human ribonuclease P gene (RP) assay to monitor nucleic acid extraction, specimen quality, and presence of reaction inhibitors. To monitor assay performance, we included positive template controls, no-template controls, and human specimen controls in all runs. When all controls exhibited expected performance, we considered a specimen to be positive for SARS-CoV-2 if all assay amplification curves crossed the threshold line within 40 cycles (C t <40). If all 3 assay results were negative, the test result was reported as negative. If all 3 assay results were not positive, an inconclusive test result was recorded and retesting was required. If both initial and retest results were inconclusive, the final result was reported as inconclusive.

Assay Efficiency
We prepared serial 10-fold dilutions of quantified RNA transcript in 10 mmol/L Tris-HCl buffer containing 50 ng/µL of yeast tRNA (Thermo Fisher Scientific) and tested them by each assay. A linear amplification was achieved over an 8-log dynamic range from 5 to 5 × 10 7 copies/reaction for all 3 assays; calculated efficiency ranged from 99.4% to 102.6% (Figure 1).

SARS-CoV-2 RNA Transcripts
We tested serial 2-fold dilutions of quantified RNA transcript prepared in buffer as above by each assay in 24 replicates/dilution. The highest dilution of transcript at which all replicates were positive was 1656 Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 26, No. 8, August 2020 defined as the limit of detection (LoD) for each assay.
The LoDs for all assays were 5 RNA transcript copies/reaction (Table 2), which is consistent with that previously demonstrated (11).

SARS-CoV-2 Genomic RNA
We tested serial half-log dilutions of SARS-CoV-2 RNA extracted from inactivated cultured virus and prepared in buffer as above in triplicate by each assay (Table 3). For all 3 assays, the LoD was ≈1 × 10 −1.5 TCID 50 /mL.

In Silico Analysis Against Available SARS-CoV-2 Genome Sequences
We evaluated the primer/probe sequences of the 3 assays against 7,158 genome sequences available from the Global Initiative on Sharing All Influenza Data (GISAID, https://www.gisaid.org) as of April 14, 2020. Nucleotide mismatches in the primer/probe regions with frequency >0.1% were sporadic among viruses (Table 6). Except for 1 nucleotide mismatch with frequency >1% (1.55%) at the eighth position of the N3 forward primer, frequency of all other mismatches was <1%. These nucleotide mismatches would not be expected to affect reaction performance. No viruses were found to have >1 mismatch in any primer/probe region.

Cross-Reactivity with Other Respiratory Pathogens and Human Microbial Flora
We evaluated the specificity of the SARS-CoV-2 rRT-PCR assay with purified nucleic acid obtained from a collection of respiratory pathogen isolates or positive clinical specimens, including human coronaviruses 229E, OC43, NL63, HKU1, SARS-CoV, and MERS-CoV (Table 7). We also tested 10 nasopharyngeal swabs samples collected in 2018 before COV-ID-19 was identified. Except for the N3 assay, which was reactive with SARS-CoV RNA as expected, we observed no false-positive results for any pathogens and specimens tested.

Specimens from Persons with Suspected Cases
Among the 2,437 clinical specimens collected from 998 persons with suspected cases for initial SARS-CoV-2 diagnostic testing, 81 (3.32%) specimens (42 nasopharyngeal, 33 oropharyngeal, 5 sputum, 1 BAL) collected from 46 persons with suspected cases were positive and 2,355 (96.64%) specimens were negative (Table 8). We did not detect SARS-CoV-2 RNA in any of the 74 serum and 10 urine specimens tested.

Laboratory-Confirmed COVID-19
Of 486 specimens serially collected from 28 persons with laboratory-confirmed COVID-19, results were SARS-CoV-2 positive for 142 (29.22%) samples (60 nasopharyngeal, 42 oropharyngeal, 13 sputum, 1 tracheal aspirate, 22 stool, and 4 serum) ( Table 8). We detected SARS-CoV-2 RNA in serum of 2 of 15 persons with laboratory-confirmed COVID-19 for whom serum was available for testing. For 1 of those casepatients, serum was collected 14 days after symptom onset and tested positive. For the other case-patient, a total of 10 serum samples were collected. Of those, specimens collected on days 9, 11, and 13 were positive; specimens collected on days 3, 19, 22, 25, and 28 were negative; and specimens collected on days 6 and 16 had inconclusive results. A total of 22 stool specimens collected from 7 case-patients were positive. We detected no SARS-CoV-2 RNA in any of the 62 urine specimens collected.

Inconclusive SARS-CoV-2 rRT-PCR Results
Inconclusive results were obtained for 40 (1.37%) of 2,923 specimens tested, including 1 (0.04%) of 2,437, from a person with a suspected case at initial testing, and 39 (8.02%) of 486 specimens collected for follow-up investigation from persons with laboratory-confirmed COVID-19. All specimens with inconclusive results had C t values >35 (Table 9). Of these, 35 (87.5%) specimens were collected >7 days after symptom onset. Times from symptom onset to collection were unknown for the other 5 (12.5%) specimens.

Quality Control Data
Among 185 no-template controls and 219 human specimen controls included with specimen testing, negative results were obtained for all (100%) controls for N1, N2, and N3 assays and expected RP values were observed for human specimen controls. C t values obtained from positive template control of 185 runs were in the expected range (data not shown).

Discussion
The COVID-19 pandemic has affected multiple countries, causing severe illness and death, with sustained and efficient person-to-person community transmission, and it continues to pose a serious public health threat. Rapid and reliable laboratory diagnosis of SARS-CoV-2 infection is a critical component of public health interventions to mitigate this threat. Our assay design and validation strategy were guided by several principles. First, we based assay designs on previous diagnostic assays that had been developed for detection of MERS-CoV (12) and SARS-CoV (10) and targeted the N gene. Because of the relative abundance of N gene subgenomic mRNA produced during virus replication (13), rRT-PCR assays targeting the N gene of coronaviruses could theoretically achieve enhanced diagnostic sensitivity. One study also showed that the N gene-based rRT-PCR assay was more sensitive than the open reading frame (ORF) 1 assay for detection of SARS-CoV-2 in clinical specimens (14). Second, we designed rRT-PCR assays on the basis of limited genetic information available soon after the emergence of SARS-CoV-2, when it was announced that a novel coronavirus of zoonotic origin was described as being similar to bat-SARS-like CoVs and only 1 SARS-CoV-2 sequence was publicly available. The N3 assay was intentionally designed to universally detect SARS-CoV-2 and other SARS-like sarbecoviruses to ensure detection of SARS-CoV-2 as this virus evolves over time and to improve early identification of future emerging novel coronaviruses from this high-risk subgenus. After completion of this study, the sequence of a new bat-SARS-like CoV, RaTG13 (EPI_SL_402131), was released on GISAID. Detected in 2013 from China, this virus appears to be the nearest bat precursor of SARS-CoV-2 (15), having 96% genome and 97% N gene sequence identity with SARS-CoV-2. All 3 assays are predicted to detect the RaTG13 strain. Third, we designed and used multiple assays for routine specimen screening to confirm virus detection when present at low concentrations and to reduce the possibility of false-negative results caused by polymorphisms within the binding sites of the target sequences, which might occur as the virus evolves. Fourth, we validated all assays by using multiple specimen types, including upper and lower respiratory specimens, serum, and stool samples, all of which have been shown to be diagnostically valuable for detection of SARS-CoV and MERS-CoV (10,16). 1660 Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 26, No. 8, August 2020  A study of SARS-CoV-2 viral load in upper respiratory tract specimens of infected patients showed that viral loads were higher in the nasopharynx than in the throat (17). Our study also showed a higher detection rate for nasopharyngeal swab samples than for oropharyngeal swab samples, although in some cases, viral load was higher in oropharyngeal than in nasopharyngeal swab samples. Our limited sputum data showed that SARS-CoV-2 seems to be more often detected and with higher viral loads in lower respiratory tract specimens than in upper respiratory tract specimens, especially later in the course of infection (72.22% positivity rate for sputum compared with 46.51% for nasopharyngeal and 30.00% for oropharyngeal samples in serial follow-up specimens). This phenomenon could be explained by the active replication of SARS-CoV-2 in the lungs,  where the SARS-CoV-2 angiotensin-converting enzyme 2 receptor predominates (18). Similar to findings for SARS-CoV (19), our results showed that the SARS-CoV-2 RNA detection rate was high in the serial follow-up stool specimens collected from casepatients with laboratory-confirmed COVID-19.
In contrast, SARS-CoV and MERS-CoV could be detected in serum/blood during the early prodromal phase of infection (12,20), whereas SARS-CoV-2 RNA was not detected in any of the serum specimens during the initial testing, although it was detected in serum collected from 2 (13.3%) case-patients >9 days after symptom onset. Similarly, whereas SARS-CoV and MERS-CoV RNA have been detected in urine (16,19), SARS-CoV-2 RNA was not detected in urine in our study nor has detection been reported (21). All specimens with inconclusive results in our study had C t values >35, with reactivity not attributable to 1 individual assay, indicating that the viral RNA levels in the specimens were at the LoD of the assay. All specimens with inconclusive results were collected either >7 days after symptom onset or from repatriates who had been quarantined on the Diamond Princess cruise ship for ≈2 weeks before specimen collection. Inconclusive results most likely resulted from low viral loads, especially in upper respiratory tract specimens collected later in the course of infection. This observation is supported by another study, which showed that SARS-CoV-2 actively replicated in the oropharynx during the first 5 days after symptoms onset only (21). If inconclusive results are obtained, collection of additional specimens and specimen types may be warranted for accurate diagnosis.
All 3 SARS-CoV-2 rRT-PCR assays proved to be both sensitive and specific with high reproducibility. The earliest specimens from US COVID-19 case-patients were confirmed by virus isolation, whole-genome or partial gene sequencing, or both (22) (GenBank accession nos. MN985325, MN988713, MN994467-8, MN997409, MT027062-4, MT039887-8, MT044257-8, MT106052-4, MT118835, MT159705-22, and MT184907-13), although not all positive followup specimens detected by rRT-PCR assays were confirmed by independent assays. An independent comparison study showed that the CDC N2 and N3 assays performed well among 7 assays targeting the N gene that were evaluated and posted by the World Health Organization (Y. Jung, unpub. data, https://www. biorxiv.org/content/10.1101/2020.02.25.964775v1). In another study that compared 7 assays, the N2 assay was shown to be among the most sensitive (23). However, both studies used enzyme chemistries not optimized by CDC for testing, and we did not observe difference in sensitivity among the 3 assays in our study.
To expedite reagent kit manufacturing by removal of the N3 assay from the panel, we analyzed results of 2,437 specimens for initial COVID-19 testing when only N1 and N2 assay results were used for interpretation. Positive and negative test results showed 100% agreement to interpretation with all 3 assays. Only 1 (0.04%) specimen with an inconclusive result, which was positive for N2 at initial testing and positive for N3 at retesting, would be reported as negative if the N3 assay was excluded from result interpretation of the CDC panel. This analysis demonstrates that interpretation of only N1 and N2 assay results agreed totally (99.96%) with the original results, and removal of the N3 assay from the panel would have a negligible effect on the ability of the test to detect positive specimens. The analytical LoD of the panel remained the same for detection of SARS-CoV-2 with or without the N3 assay included, and sensitivity of the panel was not affected. The benefits of testing with only N1 and N2 assays include simplified diagnosis of COV-ID-19 with fewer reactions for each patient specimen as well as increased testing throughput and reduced reagent cost, although removal of the N3 assay from the panel limits the ability to detect other SARS-like coronaviruses. Accordingly, the N3 assay has been removed from CDC testing interpretation, and current recommendations are to test with the N1 and N2 assays only (11). Because SARS-CoV-2 is an RNA virus with an estimated evolutionary rate of ≈1.8 × 10 -3 substitutions/site/year (24), performance of the assays included in the CDC panel will be monitored as SARS-CoV-2 continues to circulate and evolve over time. Therefore, assay designs included in the CDC panel are subject to change to account for future genetic mutations in the SARS-CoV-2 genome that may affect test sensitivity. In conclusion, the CDC rRT-PCR panel for detection of SARS-CoV-2 demonstrated high sensitivity and specificity for detecting 5 RNA copies/reaction with no observed false-positive reactivity, and it facilitated rapid detection of SARS-CoV-2 infections in humans. These assays have proven to be valuable for rapid laboratory diagnosis and support, clinical management, and infection prevention and control of COVID-19.