key: cord-0772234-x03cb307
authors: Li, Jingyuan; Lin, Weishi; Du, Pibo; Liu, Wei; Liu, Xiong; Yang, Chaojie; Jia, Ruizhong; Wang, Yong; Chen, Yong; Jia, Leili; Han, Li; Tan, Weilong; Liu, Nan; Du, Junjie; Ke, Yuehua; Wang, Changjun
title: Comparison of reverse-transcription qPCR and droplet digital PCR for the detection of SARS-CoV-2 in clinical specimens of hospitalized patients
date: 2022-03-19
journal: Diagn Microbiol Infect Dis
DOI: 10.1016/j.diagmicrobio.2022.115677
sha: 462a627abf4bdfbccb63a223080d07f3a3c03f05
doc_id: 772234
cord_uid: x03cb307

Accurate detection of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is not only necessary for viral load monitoring to optimize treatment in hospitalized coronavirus disease (COVID-19) patients, but also critical for deciding whether the patient could be discharged without any risk of viral shedding. Digital droplet PCR (ddPCR) is more sensitive than reverse transcription quantitative real-time polymerase chain reaction (RT-qPCR) and is usually considered the superior choice. In the current study, we compared the clinical performance of RT-qPCR and ddPCR using oropharyngeal swab samples from patients hospitalized in the temporary Huoshenshan Hospital, Wuhan, Hubei, China. Results demonstrated that ddPCR was indeed more sensitive than RT-qPCR. Negative results might be caused by poor sampling technique or recovered patients, as the range of viral load in these patients varied significantly. In addition, both methods were highly correlated in terms of their ability to detect all three target genes as well as the ratio of copies of viral genes to that of the IC gene. Furthermore, our results evidenced that both methods detected the N gene more easily than the ORF gene. Taken together, these findings imply that the use of ddPCR, as an alternative to RT-qPCR, is necessary for the accurate diagnosis of hospitalized COVID-19 patients.

Since its outbreak in December 2019, coronavirus disease 2019 (COVID-19) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has spread rapidly and resulted in a worldwide pandemichttps://coronavirus.jhu.edu. In this situation, there is an unprecedented need for diagnostic nucleic acid testing (Binnicker, 2020 , Kilic et al., 2020 .

The availability and reliability of rapid nucleic acid tests facilitates the quick identification of infected individuals. This mobilizes the appropriate utilization of scarce public health and medical resources, including those related to contact-tracing, isolation, personal protective equipment, and therapeutic devices (Rogers et al., 2020) .

Reverse transcription quantitative real-time polymerase chain reaction (RT-qPCR) is considered the gold standard method for the diagnosis of SARS-CoV-2 infection and is routinely used in the epidemiological screening of individuals with suspected COVID-19.

However, for hospitalized patients, the sensitivity of RT-qPCR is usually not sufficient to discriminate positive patients with a very low SARS-CoV-2 viral load, from recovered patients without risk of viral shedding (Tahamtan and Ardebili, 2020 , Veyer et al., 2020 , Yu et al., 2020 . In addition, the inability to quantify the viral load in patients renders it impossible to monitor viral load changes during treatment (Fajnzylber et al., 2020 , Wolfel et al., 2020 , Yu et al., 2020 . Therefore, more sensitive and robust detection methods, especially for low and residual viral load samples, are required for accurate SARS-CoV-2 diagnosis and to mitigate the shortcomings of RT-qPCR. Fortunately, digital droplet PCR (ddPCR) provides absolute quantification through Poisson statistics after limiting dilution and endpoint PCR, offering a more sensitive method than RT-qPCR (Milosevic et al., 2018, Salipante and Jerome, 2020) . ddPCR has been used to detect small fold changes in copy number variation or gene expression, and rare mutations in cancer diagnostics and infectious disease diagnostics, including dengue virus, hepatitis A virus, norovirus, and SARS-CoV-2 (Abachin et al., 2018 , Coudray-Meunier et al., 2015 , Persson et al., 2018 .

In hospitalized patients with suspected COVID-19, relying on RT-qPCR to confirm SARS-CoV-2 infection, the accurate detection and precise diagnosis of viral content are helpful and necessary for viral load monitoring and treatment adjustment (Fajnzylber et al., 2020 , Wolfel et al., 2020 . Additionally, accurate diagnostic techniques are critical for deciding whether the patient has recovered without risk of viral shedding and can be discharged, or the patient is still shedding viral debris and requires further treatment , Wolfel et al., 2020 . In the present study, we compared the clinical performance of RT-qPCR and ddPCR in patients hospitalized at Huoshenshan Hospital, a temporary hospital built in response to the COVID-19 outbreak in Wuhan, China (Zhihui. et al., 2020) . Furthermore, we assessed the correlation between these two methods in determining RNA copies of different viral genes from different individuals. Our results provide a unique characterization of SARS-CoV-2 viral load in clinical specimens from hospitalized patients who underwent treatment or were awaiting discharge. Additionally, our findings demonstrate a notable correlation between SARS-CoV-2 detection using RT-qPCR and ddPCR and might indicate viral load fluctuations within individuals and the divergence of genome replication within different regions.

The Huoshenshan Hospital was specifically established to treat COVID-19 patients in response to the outbreak in Wuhan, Hubei, China (Zhihui. et al., 2020) . All specimens used in this study were obtained from patients with confirmed SARS-CoV-2 infection before they were transferred to Huoshenshan Hospital. In total, more than 4,000 patients were admitted and treated in this hospital (Zhihui. et al., 2020) . All recovered patients must test SARS-CoV-2-negative for two consecutive days before being discharged. Some of these patients might have recovered from the infection during the sampling; therefore, the samples obtained from them tested negative for SARS-CoV-2 and were used as negative controls for our experiments. Oropharyngeal swabs were collected under aseptic conditions and placed in sterile tubes containing viral transport medium. Samples were stored at 4°C and transported directly to the diagnostic laboratory for further examination.

Prior to RNA extraction, all samples were treated at 56°C for 30 min to inactivate SARS-CoV-2. Within 2 h of inactivation, total viral RNA was extracted from the supernatant using Prefilled Viral Total NA Kit-Flex (KFRPF-805296; Fisher Scientific, LOCATION) following the manufacturer's instructions. Briefly, 200 μL specimens were used for extraction, and elution was set to 50 μL. Extracted RNA samples were either immediately subjected to RT-qPCR and ddPCR protocol or stored at -80°C for further studies. 

RT-qPCR was conducted using the SARS-CoV-2 RNA detection kit (Sansure Co., Ltd., Changsha, China) with the SLAN Real-time PCR System following the manufacturer's instructions. Briefly, the total volume of the reaction mixture was 25 μL and it contained 10 μL RNA template. The reaction conditions were as follows: reverse transcription at 50°C for 30 min; cDNA pre-denaturation at 95°C for 1 min; denaturation at 95°C for 15 s (45 cycles), then annealing and elongation (with fluorescence monitoring) at 60°C for 30 s, and a final step at 25 °C for 10 min. A cycle threshold (CT) value ≤ 40 indicated a positive result and that >40 represented a negative result. Samples positive for all three genes were considered positive, whereas samples positive for the IC gene and only one viral gene indicated suspected COVID-19 cases. However, samples positive for the IC gene and negative for both viral genes were considered negative. Samples that tested negative for the IC gene were considered defective.

Workflow procedures for ddPCR were performed according to the manufacturer's instructions for the RainSure DropX-2000 Droplet Digital PCR System using the RainSure Novel Coronavirus (SARS-CoV-2) Nucleic Acid Detection Kit [19] . Briefly, 25 μL reaction mixture contained 10 μL SARS-CoV-2 one-step RT ddPCR master mix, 4 µL enzyme mix, and 10 μL RNA extracted from patient samples. Firstly, 70 μL droplet generation oil and 25 μL reaction mixture were loaded into an oil well and a sample well, respectively. Thereafter, a gasket with filters was mounted onto the wells of the reagentloaded cartridges. The cartridges were then loaded into the instrument and the droplet generation process automatically commenced using the following thermal cycling protocol: 49°C for 20 min (reverse transcription); 97°C for 12 min (DNA polymerase activation); 40 cycles at 95.3°C for 20 s (denaturation), and then 52°C for 1 min (annealing); and finally 20°C (cooling) for infinite hold. The cartridges were then transferred and loaded onto the DScanner-2000 for multi-channel fluorescence detection of droplets. Results were interpreted in a similar manner to those of RT-qPCR.

Analysis of ddPCR data was performed using the analysis software GeneCount V1.60b0318 (RainSure Scientific). Concentrations of target RNA sequences along with their Poisson-based 95% confidence intervals were provided by the software. The Mann-Whitney test was performed to make comparisons between two groups. Additionally, the Spearman rank correlation test was used to analyze the correlation between the CT values of RT-qPCR and log2 values of copies determined by ddPCR. Computations were performed using R software version 3.6. P values < 0.05 were considered statistically significant.

Among the 130 clinical samples obtained from hospitalized patients, 89, 9, and 32 samples tested positive, suspected, and negative for COVID-19, respectively, using RT-qPCR.

Conversely, ddPCR detected 93, 21, and 16 positive, suspected, and negative COVID-19 samples, respectively. Although both methods are considered to have a high specificity, ddPCR demonstrated a higher true positive rate and sensitivity than RT-qPCR (Table 1) .

Furthermore, both methods successfully detected the IC gene in all 130 samples. Regarding the ORF gene, all 89 positive samples detected with RT-qPCR also tested positive using ddPCR (100%). However, 6 of the 41 samples that tested negative for the ORF gene using RT-qPCR, came back positive using ddPCR (14.6%). Moreover, the 98 samples that tested positive for the N gene using RT-qPCR were also positive using ddPCR (100%), although 14

of the 32 samples that tested negative for the N gene using RT-qPCR, tested positive using ddPCR (43.75%) ( Table 2 ). These results indicate that ddPCR might be more sensitive than RT-qPCR for detecting the ORF and N genes, although both methods successfully detected the IC gene similarly; this was consistent with the results from previous studies (Falzone et al., 2020 , Rao et al., 2020 , Veyer et al., 2020 . Regarding CT values determined using RT-qPCR, CT ORF values were higher than CT N values in all 89 positive samples. However, 8 samples (8/89; 8.98%) determined using ddPCR demonstrated a higher value ORF than value N , which might be due to the higher accuracy with which ddPCR detects viral targets (Supplement Table 1 ) (Chen et al., 2020) . respectively. The number of copies of the IC, ORF, and N genes ranged from 0.02 to 9425.00 (median: of 146.10), 0.12 to 7727.00 (median: 6.42), and from 0.14 to 13561.00 (median: of 8.814), respectively, when using ddPCR. For both RT-qPCR and ddPCR techniques, the median CT value of the IC gene was higher in the COVID-19-negative group than that in the COVID-19-positive group ( Figure 1A and B) . Moreover, higher CT values of the IC gene in negative samples indicated lower total RNA content compared to positive samples. This might be due to poor sampling techniques, which result in lower viral content collected on oropharyngeal swabs.

The median CT values of the ORF gene were higher than those of the N gene, detected using RT-qPCR (Rao et al., 2020) ; however, there were no obvious differences noted when using ddPCR ( Figure 1C and D) . This suggests that further study is required regarding individual differences of these two methods.

Considering that RT-qPCR and ddPCR are semi-quantitative and quantitative detection methods, respectively, we aimed to explore whether any correlation existed between these two methods. Surprisingly, for all 89 COVID-19-positive samples detected in several different batches, we identified that the results for all three target genes demonstrated a high correlation between RT-qPCR and ddPCR techniques. Particularly, R 2 values of 0.9752, 0.8455, and 0.8942 were calculated for the IC, ORF, and N genes, respectively (Figure 2A, 2B and 2C ).

These results indicate that although the CT value determined using RT-qPCR is considered relatively quantitative, R 2 values are more accurate for determining the absolute content of the target gene (i.e., viral load) in detecting SARS-CoV-2. Therefore, contrary to prior belief that only valuable qualitative results were obtained from RT-qPCR, we might be able to compare the viral load of different samples using their CT value determined using RT-qPCR (Yu et al., 2020) .

A single SARS-CoV-2 genome only contains one copy of both the ORF and N genes.

This suggests that copies of the ORF gene should be equal to those of the N gene (Kim et al., 2020) . However, RT-qPCR-revealed CT values of the ORF gene were higher than those of the N gene in most samples. These results suggest that the N gene consisted of more nucleic acids than the ORF gene within the same sample, where amplification efficiency was not considered.

Moreover, ddPCR also detected that most of the copy values of the ORF gene were lower than those of the N gene at the individual level ( Figure 2D and 2E) (Chen et al., 2020) . Interestingly, the ddPCR method detected no significant differences between the overall values of the ORF and N genes ( Figure 2D ). This might be due to a few exceptions in which outlying results misrepresented the overall effects, indicating that a more detailed comparison is necessary for analyzing absolute quantitative results obtained with ddPCR.

The CT values of the ORF and N genes demonstrate a strong correlation, using both RT-qPCR (R 2 = 0.8897) and ddPCR detection methods (R 2 = 0.9097), as presented in Figure 2F and 2G. The correlation between the ORF and N genes in any one method elucidated that these two target genes, within the viral genome, are closely related (Chen et al., 2020 , Yu et al., 2020 . In general, both methods were more sensitive for the detection of the N gene than the ORF gene (Chen et al., 2020) .

As described above, we confirmed that the values of the ORF and N genes exhibited a linear correlation in any one detection method. Moreover, the detection of all three target genes also had a linear correlation across both methods. Therefore, we pondered whether the ratio between any two target genes was correlated within the same method or between the two methods. To further determine the correlation of the ratio between any two targets, we used CT IC / CT ORF , CT IC / CT N , and CT ORF / CT N for RT-qPCR and log2(value IC /value ORF ), log2(value IC /value N ), and log2(value ORF /value N ) for ddPCR.

We identified that CT IC / CT ORF and CT IC / CT N demonstrated a very strong linear correlation in the RT-qPCR method, with a similar result verified with ddPCR ( Figure 3A and B). Moreover, a linear correlation was detected between CT IC / CT ORF in RT-qPCR and log2(value IC /value ORF ) in ddPCR, with similar results obtained regarding these ratios for the N gene ( Figure 3C and D) . Unsurprisingly, we did not observe any correlation between CT ORF / CT N in RT-qPCR and log2(value ORF /value N ) in ddPCR ( Figure 3E ). The ratio of copies of viral genes compared to that of the IC revealed that individuals have fluctuating viral loads to some extent, and a very strong correlation within and between the two methods. This indicates that the analysis of these ratios is reliable and could efficiently reflect variable virus loads during infection.

Positive specimens, obtained by screening for SARS-CoV-2 infection among suspected COVID-19 cases, were sampled from patients during the early phase of infection. Therefore, the qualitative observation of viral infection was generally sufficient to help make correct medical decisions (Binnicker, 2020) . For hospitalized COVID-19 patients, nucleic acid testing for the detection of viral RNA is usually required when patients have obviously recovered and are awaiting discharge, or when patients experience worsening of symptoms and require treatment optimization (Wolfel et al., 2020 , Yu et al., 2020 . Therefore, it may be possible to characterize viral load fluctuation in these patients (Fajnzylber et al., 2020 , Wolfel et al., 2020 , Yu et al., 2020 . The results of RT-qPCR and ddPCR performed using specimens collected from hospitalized COVID-19 patients demonstrate that ddPCR is indeed more sensitive than RT-qPCR. Moreover, negative specimens might be the result of poor sampling techniques, as the viral load in these patients varied significantly (Dang et al., 2020 , de Kock et al., 2020 , Falzone et al., 2020 , Muenchhoff et al., 2020 , Rao et al., 2020 .

Both methods demonstrated a strong correlation regarding the detection of all three target genes, and between the ratio of the copies of viral genes to that of the IC gene. These results indicate that the CT values of genes detected using RT-qPCR could be used to evaluate the number of viral copies in a specimen when it was not possible to utilize ddPCR to directly determine viral copies (Chen et al., 2020 , Yu et al., 2020 . Furthermore, our results evidence that the N gene was more easily detected by both methods, which might be due to the efficiency of amplification and fluorescence monitoring of these two genes as well as their primers and probes (Chen et al., 2020 , Wolfel et al., 2020 . This might indicate a distinction between primers, probes, polymerase, and reaction mixtures, or imply deviations of genome replication or stability within the N and ORF regions, which requires further study (Rao et al., 2020) .

In summary, we compared the clinical performance of RT-qPCR and ddPCR in detecting SARS-CoV-2 infection in oropharyngeal swab samples from hospitalized patients.

Results indicated that ddPCR is indeed more sensitive than RT-qPCR. Additionally, COVID-19-negative specimens might result from poor sampling techniques, as the viral load in these patients varies significantly. All three target genes and their ratios demonstrated strong correlations between both methods; however, both methods were more sensitive for detecting the N gene than the ORF gene. Our results evidence that ddPCR should be used as an alternative to RT-qPCR for the accurate diagnosis of COVID-19 in hospitalized patients.

The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. Correlation between −log2(value IC / value ORF ) and -log2(value IC / value N ) determined using ddPCR. (C)

Correlation between CT IC / CT ORF and −log2(value IC / value ORF ) determined by RT-qPCR and ddPCR, respectively. (D) Correlation between CT IC / CT N and −log2(value IC /value N ) determined by RT-qPCR and ddPCR, respectively. (E) Correlation between CT N / CT ORF and −log2(value N /value ORF ) determined by RT-qPCR and ddPCR, respectively.

Comparison of reversetranscriptase qPCR and droplet digital PCR for the quantification of dengue virus nucleic acid

Challenges and Controversies to Testing for COVID-19

Influence of Different Inactivation Methods on Severe Acute Respiratory Syndrome Coronavirus 2 RNA Copy Number

A comparative study of digital RT-PCR and RT-qPCR for quantification of Hepatitis A virus and Norovirus in lettuce and water samples

Comparison of qualitative and quantitative analyses of COVID-19 clinical samples

Sensitive detection and quantification of SARS-CoV-2 by multiplex droplet digital RT-PCR

Molecular and Immunological Diagnostic Tests of COVID-19: Current Status and Challenges

The Architecture of SARS-CoV-2

Analytical comparisons of SARS-COV-2 detection by qRT-PCR and ddPCR with multiple primer/probe sets

Applying Standard Clinical Chemistry Assay Validation to Droplet Digital PCR Quantitative Liquid Biopsy Testing

Multicentre comparison of quantitative PCR-based assays to detect SARS-CoV-2

Comparison between RT droplet digital PCR and RT real-time PCR for quantification of noroviruses in oysters

Comparing nasopharyngeal swab and early morning saliva for the identification of SARS-CoV-2

Digital PCR-An Emerging Technology with Broad Applications in Microbiology

ddPCR: a more accurate tool for SARS-CoV-2 detection in low viral load specimens

Real-time RT-PCR in COVID-19 detection: issues affecting the results

Temporal profiles of viral load in posterior oropharyngeal saliva samples and serum antibody responses during infection by SARS-CoV-2: an observational cohort study

Highly sensitive quantification of plasma SARS-CoV-2 RNA shelds light on its potential clinical value

Virological assessment of hospitalized patients with COVID-2019

Quantitative Detection and Viral Load Analysis of SARS-CoV-2 in Infected Patients

Specialized Hospital Construction and Management in Wuhan(in Chinese)

Weishi Lin: Methodology, Writing original draft, Funding acquisition

Pibo Du: Conceptualization, Writing original draft, Investigation, Formal analysis

Resources, Validation

Resources, Validation

Chaojie Yang: Resources, Validation. Ruizhong Jia: Resources, Validation

Resources, Validation

Resources, Validation. Leili Jia: Resources, Validation

Investigation, Visualization. Weilong Tan: Software, Data curation

Investigation, Visualization

Project administration, Writing -review & editing

Project administration, Writing -review & editing, Funding acquisition

Project administration, Writing -review & editing, Funding acquisition

The authors declare that they have no competing interests.