On the nature of false-negative results in the detection of SARS-CoV-2 coronavirus by nucleic acid amplification methods

K.T. MOMYNALIEV, Dr. of Biological Science, docent, expert, dhoroshun@gmail.com

I.V. IVANOV, Candidate of Medical Science, CEO, ivi1976@bk.ru

For quoting: Momynaliev K.T., Ivanov I.V. On the nature of false-negative results in the detection of SARS-CoV-2 coronavirus by nucleic acid amplification methods. – Vestnik Roszdravnadzora». – 2020. – № 2. – P. 11–19.

DOI: https://doi.org/10.35576/2070-7940-2020-2-11-19

On the nature of false-negative results in the detection of SARS-CoV-2 coronavirus by nucleic acid amplification methods

Key words: COVID-19; SARS-CoV-2, real-time reverse transcription polymerase chain reaction (RT-PCR), false-negative results, diagnostic sensitivity, diagnostic specificity.

Introduction

The 2019 pandemic, caused by coronavirus of severe acute respiratory syndrome-2 (SARS-CoV-2) has become a global public health threat. The diagnosis of coronavirus disease 2019 (COVID-19) is based on virus isolation or a positive polymerase chain reaction (PCR) result from sputum, nasal swab or pharyngeal swab, but false-negative results may occur. This review discusses the main reasons for false-negative results, when RT-PCR is used to detect SARS-CoV-2 coronavirus. Among the main reasons preanalytical errors (sample collection, processing, transportation, storage and preparation for testing), variability in the predictive value of the test from the onset of symptoms, and the bioavailability of SARS-CoV-2 in different types of clinical samples of patients.

During the development of the coronavirus infection pandemic in the Russian Federation, an important task is to test the population for the early detection of patients diagnosed with COVID-19. To solve this problem, a sufficient number of tests are required that can detect the virus in patients. Federal State Organization «National quality institute» of Roszdravnadzor is an expert organization, that examines the quality, effectiveness and safety of medical devices for the purpose of their state registration at the instruction of Roszdravnadzor. Evaluation of the claimed test-systems, their sensitivity and specificity, cross-reactivity are key characteristics when considering the submitted documents. Studying the experience of other countries when working with various test-systems is an important component of the expert’s activities.

The aim of the article

To examine the reasons of false-negative results when using tests for detecting SARS- CoV-2 coronavirus by nucleic acid amplification methods.

The problem of false-negative tests for PCR diagnostics of SARS-CoV-2

On December 31, 2019 there were reports of cases of pneumonia of unknown origin in Wuhan, China. On January 7, 2020 it was established that the causative agent was a new coronavirus, which received a temporary designation «2019 -nCoV». On January 30, 2020 the WHO Director-General announced, that the outbreak of new coronavirus (2019-nCoV) infection represented a public health emergency of international concern. On February 11, 2020 WHO officially designated the infection caused by a new coronavirus as COVID-19 («Coronavirus disease 2019»). On February 11, 2020, the International Committee on Taxonomy of Viruses assigned the official name to the infectious agent – SARS-CoV-2.

To confirm the clinical diagnosis of COVID-19 a nucleic acid amplification method is used (Real-time reverse transcription polymerase chain reaction (RT-PCR) or isothermal reverse transcriptional amplification (RT-IA))1. Nucleic acid amplification tests, as a rule, have high diagnostic specificity and sensitivity2 (over 95%). This means that more than 95% of clinical samples containing SARS-CoV-2 will be detected by the amplification of nucleic acid method and in 95 % of cases the virus will not be detected if it is absent or there are other pathogens of acute respiratory infections.

The use of tests based on nucleic acids amplification method in Russian Federation has sharply increased lately. The number of tests on coronavirus infection conducted in Russia exceeded 1,5 million, from 80 to 90 thousand tests are performed daily. Along with the widespread introduction of tests into clinical practice there are reports that the real sensitivity of PCR method is in the range of 60–80%, that the clinical picture of COVID-19 is revealed in patients with negative results of the PCR test [8]. For example, Feng H. and co-author (2020) informed of a patient whose diagnosis of COVID-19 was not confirmed on the basis of four successive pharyngeal swab RT-PCR tests, while a computer tomography of the lungs showed a patchy opacity of frosted glass on admission and that progressed rapidly [5]. Only the fifth RT-PCR test on the fifth day after the patient’s admission, had finally showed positive results.

Apart from the fact that false-negative results can delay timely treatment of patients with coronavirus infection, there is a threat to public health due to the spread of infection from patients with such research results. For example, people with false-negative results can loosen the social distance and other measures, supposed to reduce the transmission of the virus to others, and clinicians with such result will inadvertently transmit the virus to patients and colleagues in their professional activity.

In this review we attempted to understand the nature of false-negative results of PCR tests in laboratory diagnostics and evaluate measures that will reduce them.

Preanalytical errors

At present time, there are incontestable evidences that preanalytical phase is a main source of errors in laboratory research, when used for diagnostic or research purposes [9, 11].

The safety and quality of RT-PCR testing can be compromised by the absence or incorrect identification of a patient and/or sample, the collection of inappropriate or inadequate materials (in terms of quality or volume), improper transport and storage conditions of a sample (for example, unstable temperature, prolonged transport time), the presence of interfering substances (for example, release of cellular components that could interfere the analysis due to the whole blood freezing, use of inappropriate additives), and a number of procedural technological problems, occurring during the preparation of sample, including, but not limited to the pipetting errors in manual preparation of sample or aliquoting, sample cross-contamination (tabl.). Key specific issues, that can affect the quality of RT-PCR analysis, include sample contamination (even trace amounts of external DNA can compromise test results) and testing conducted on patients receiving antiretroviral therapy, which can then lead to false-negative results3.

Sensitivity of tests as one of the sources of false-negative results.

Currently a number of tests based on amplification of nucleic acids are registered in Russian Federation. Generally, diagnostic sensitivity (DS) characteristics are reported in the format: DS, determined in clinical trials (95% of CI: lower limit – upper limit)4, for example, 100% (95%: 95,6 – 100%). The width of the confidence interval depends on the number of bio samples that were used in clinical trials. The more samples, the smaller the confidence interval, and vice versa.

Diagnostic sensitivity of all PCR-tests determined during clinical trials, is 100% which means that all samples containing SARS-CoV-2 were detected by this PCR-test. However, when these results are extrapolated to the population the test sensitivity is expected to decrease and no longer reaches 100%, for example, the average value may be 95% with a confidence level of 95%.

What threats arise for the general population can be illustrated by the examples of Spain and the USA, for example, if the sensitivity of the PCR-test is 95% [15].

The mayor of Madrid has assumed that 80% of 6,5 million of citizens of Madrid will be infected with COVID-19. If the entire population was tested out of expected 5,2 million of infected, 260000 people would be mistakenly classified as free of infection. Even with less common or targeted testing among random samples the number of false-negative tests can be enormous. Likewise, it has been estimated that California’s COVID-19 prevalence could exceed 50% by mid-May 2020.

With a population of 40 million, one million of false-negative results could be expected with extensive testing. Even if only 1% of the population were tested, 10,000 false negative results could be expected. If COVID-19 infection rate among 4 million doctors, nurses and other healthcare workers providing direct patient care in the United States of America, is even 10% (well below the majority of national propagation prediction), more than 20000 false-negative results can be expected, if every doctor is tested. If the sensitivity of the test was only 70%, as indicated in early reports, the number of false-negative would be over 100000 [4]. Regardless of the accuracy of these numbers, each of these healthcare workers might be the source of disease dissemination, despite the apparent confidence in the negative result of COVID-19 test.

It should be noted that all registered PCR test-systems have limitations – the detection limit (minimum detectable concentration) of the virus in the sample. In practice, this means that if the detection limit of PCR test-system is 1 000 copies of SARS-CoV-2 per 1 ml of sample, then all samples with a concentration of 1000 copies/will be missed by this test and the testing result will be negative.

One more reason of false-negative results that has recently been identified is the risk of active recombination and mutation which are connected with the error-prone RNA-dependent RNA polymerase of coronaviruses. Shen and co-authors found a significant level of viral diversity in some infected patients, on average intra-individual viral variants indicating the rapid evolution of SARS-CoV-2 [13].

In other study Yi discovered five different SARS-CoV-2 haplotypes, which usually reflect active genetic recombination [18]. This virus evolution not only explains the heterogeneity observed in intra-individual immune response, virulence, pathogenicity and transmissibility, but can also compromise the accuracy of RT-PCR determination.

Detection of SARS-CoV-2 in different types of clinical samples

The main type of biomaterial for laboratory research is the material obtained by taking a nasopharynx and/or oropharynx swab. As an additional material for research sputum (if any), bronchial washings obtained by fibro-optic bronchoscopy (bronchoalveolar lavage), (endo-)tracheal, nasopharyngeal aspirate, biopsic or autopsy material of lungs, whole blood, serum, blood plasma, feces can be used.

Obviously the biodistribution of SARS-CoV-2 in various tissues of patients with coronavirus disease 2019 (COVID-19) is important in identifying the SARS-CoV-2 coronavirus, that is, it is necessary to understand which and when biomaterial should be tested in a patient if infection is suspected. Wang and co-authors examined the samples of patients based on clinical indications [14]. Throat swabs were taken from the majority of patients in 1–3 days after hospitalization. Samples of blood, sputum, feces, urine and nasal discharge were collected throughout the illness. Bronchoalveolar lavage and biomaterial obtained by fibro-optic bronchoscopy were taken from the patients with severe disease or after artificial lung ventilation. Cycle therehold (Ct) less than 40 was interpreted positive to SARS-CoV-2 RNA. A total of 1 070 samples from 205 patients with COVID-19 were examined, middle age 44 years old (ranged 5–67 years old), 68% of men. Most of the patients had such clinical implications as fever, dry cough; 19% of patients had severe state of disease.

In the course of conducted work the following results were obtained. Bronchoalveolar lavage fluid samples showed the highest number of positive results (14 out of 15; 93%), then sputum (72 из 104; 72%), nasal swab (5 out of 8; 63%), obtained by fibro-optic bronchoscopy (6 out of 13; 46%), throat swab (126 out of 398; 32%), feces (44 out of 153; 29%) and blood (3 out of 307; 1%). None of the 72 urine samples has given positive testing results. Average cycle thereholds for all sample types were over 30 (<2,6 104 copies/ml), except for nasal swabs with an average cycle therehold of 24,3 (1,4 106 copies/ml) indicating higher viral loads.

Twenty patients had from 2 to 6 samples, collected at the same time. Viral RNA was detected in single samples from 6 patients (respiratory samples, feces and blood), while 7 patients discharged the samples of respiratory tract of the and in the feces (n = 5) or blood (n = 2). Live SARS-CoV-2 was observed in a stool sample of 2 patients who did not have diarrhea.

In this research SARS-CoV-2 was detected in samples from several sites of patients with COVID-19, with samples from the lower respiratory tract most often testing positive for the virus.

Apparently, testing samples from multiple sites can improve sensitivity and reduce false-negative test results.

Variations of false-negative results of SARS-CoV-2 PCR-tests in time from the beginning of infection

There is a concept of «window period», in which testing is most likely to show false-negative test results, for example, if the virus concentration is still below the detection limit of the PCR-test. If negative tests conducted within the window period are considered as conclusive evidence that the sample does not contain the SARS-CoV-2 coronavirus, then a preventable transmission can occur.

Therefore, it is very important to understand how the predictive value of a test changes over time from infection and symptom onset in order to avoid false-negative tests in the early stages of infection.

The work of Kucirka L. and co-author evaluated false-negative results by days from the moment of infection [6]. The authors calculated post-test probability of infection according to the prior probability based on the frequency of infection in close household contacts with SARS-CoV-2 in Shenzhen (77/686, 11,2%) [3]. The authors supposed that specificity for RT-PCR is 100%.

As a result, it was shown that in four days of infection before the typical time of the onset of symptoms (day 5) the probability of false-negative test of an infected person decreases from 100% on the first day (95% of CI 69-100%) to 61% on the fourth day (95% of CI 18-98%), although there is a significant uncertainty in these numbers.

On the day of symptom onset, the median of false-negative results was 39% (95% of CI 16-77%) (pic. 1, upper panel). Then it decreased to 26% (95% of CI 18-34%) on the eighth day (in 3 days after symptoms onset), and then started growing again, from 27% (95% of CI 20-34%) on the nineth day to 61% (95% of CI 54-67%) on the twenty first day.

Thus, according to Kucirka L. And co-author the tests conducted on the first day of symptoms onset are more informative.

It is also obvious, that the amount of virus in a tampon is likely to vary between patients, местом probe sampling site (nasal, throat or sputum) and time as the infection progresses. From this point of view, a study of Wikramaratna and co-author is of interest, in which the researchers also tried to evaluate the probability of false-negative depending on the number of days between symptoms onset and the test and how this relates to the sites from where the smears were taken [12].

The study used data from the researches that reported the results of PCR-tests of hospital patients who tested positive for COVID-19 at least once [7, 16, 19]. A total of 426 tests were obtained from 39 patients in 3 cohorts. However, only two studies reported on the site of the swab (nasal or throat) for each individual test; therefore, the analysis was limited with this data i.e., 298 tests from 30 patients

(150 nasal swabs and 148 of throat). Data was analyzed using binomially distributed generalized additive mixed models (GAMM) with the package of mgcv in a statistical software application R.

In the course of this study, the following results were obtained. Throat swabs straight after the onset of symptoms are 6,39% likely to be positive, rather than a nasal swab. The probability of a positive test decreases with the number of days after the onset of symptoms; for nasal smear, the percentage probability of a positive test decreases from 94,39% (86,88, 97,73) at manifest (0 days) to 67,15% (53,05, 78,85) on the tenth day.

By day 31 there is only 2,38% (0,60, 9,13) of a positive result probability (numbers for nasal swabs: 88% (75,18, 94,62), 47,11% (32,91, 61,64) and 1,05% (0,24, 4,44) for 0, 10 and 31 days respectively. The data obtained are comparable to the model proposed by Kucirka L. and co-authors.

Thus, the probability of a false negative test result depends on the number of days since the onset of symptoms. This means that simple reports of positive and negative test results among individuals who are tested only once will underestimate the true number of positive results in that group.

Picture 2 shows, how changing the shape and speed of the gamma distribution affects the average rate of false-negative results. If infected individuals are tested late, then the rate of false-negative results is 4 times higher in this group then of patients who are tested early. The probability of misidentifying the patient’s status is greatly decreased due to false-negative test, if all negative tests are repeated in 24 hours. In this scenario a realized error frequency (actual rate of false-negative tests) will be proportional to the underlying prevalence of infection; only if everyone in the group is infected, the probability of a false negation will be equal to the proportion of negative tests (since there will be no true negative results from uninfected people).

The retrieved data confirms the need to exercise caution when interpreting SARS-CoV-2 PCR-tests, especially if they are conducted at early stages of infection. If there are serious clinical suspicious, patients cannot be excluded only on the basis on PCR-test and the clinical and epidemiological situation should also be carefully studied. In many cases the exposure time is unknown, and testing is conducted based on the time of the onset of the symptoms.

The authors advise to test sputum samples whenever possible, since they provide a higher viral detection level then a nasopharyngeal swab.

Conclusion

The emergence of controversial results between chest CT and RT-PCR described in some studies, along with the evidences that the virus shedding may still occur at undetectable levels in the very early and late phases of SARSCoV-2 infection indicate that the RT-PCR test results should always be interpreted in a broader context. The data resulting from the preliminary studies demonstrate that asymptomatic (subclinical) patients with COVID-19 can demonstrate very early, but paradigmatic CT changes even prior to positive RT-PCR also confirms that the most effective strategy for diagnosing COVID-19 of suspected patients is a combination of two highly sensitive approaches: SARS-CoV-2 RT-PCR with clinical and epidemiologic data and the chest CT results [1, 2, 17].

In addition to it, it is likely that the external quality assessment must be established as soon as possible for quality monitoring processes along with the implementation of internal quality control and safety of the medical activity of the laboratory service. In spite of striving to provide a high throughput and short processing time of studies for SARS-CoV-2 diagnosing, extensive validation of RT-PCR analysis is urgently needed to enable the most appropriate public health approaches to be taken on an individual and population basis.

Finally, in addition to RT-PCR tests apart from their diagnostic sensitivity and specificity it is necessary to indicate the information about their predictive value: positive and negative predictive value5. This information will enable the doctors to carefully interpret the negative tests.

RESOURCES  

  1. Ai T., Yang Z., Hou H., Zhan C., et al. Correlation of chest CT and RT-PCR testing in coronavirus disease 2019 (COVID- 19) in China: a report of 1014 cases. Radiology 2020:200642. DOI: 10.1148/ radiol.2020200642.
  2. Backer J.A., Klinkenberg D., Wallinga J. Incubation period of 2019 novel coronavirus (2019-nCoV) infections among travellers from Wuhan, China, 2028 January 2020. Euro Surveill 2020;25. DOI: 10.2807/1560-7917.ES.2020.25.5.2000062.
  3. Bi Q., Wu Y., Mei S., Ye C., Zou X., Zhang Z., et al. Epidemiology and Transmission of COVID-19 in Shenzhen China: Analysis of 391 cases and 1,286 of their close contacts http://dx.doi.org/10. 1101/2020.03.03.20028423.
  4. Fang Y., Zhang H., Xie J., et al. Sensitivity of chest CT for COVID-19: comparison to RT-PCR. Radiology. Published online Feb 19 2020; https://doi.org/10.1148/radiol.2020200432.
  5. Feng Hao, Yujian Liu, Minli Lv, and Jianquan Zhong. A case report of COVID-19 with false negative RT-PCR test: necessity of chest CT. Jpn J Radiol. 2020 Apr 7: 1–2. doi: 10.1007/s11604-020-00967-9.
  6. Kucirka L., Lauer S., Laeyendecker O., medRxiv. DOI: https://doi. org/10.1101/2020.04.07.20051474.
  7. Kujawski S..A, Wong K.K., Collins J.P., et al. First 12 patients with coronavirus disease 2019 (COVID-19) in the United States. DOI: 10.1101/2020.03.09.20032896.
  8. Li D., Wang D., Dong J., et al. False-Negative Results of Real- Time Reverse-Transcriptase Polymerase Chain Reaction for Severe Acute Respiratory Syndrome Coronavirus 2: Role of Deep- Learning-Based CT Diagnosis and Insights from Two Cases Korean J Radiol. 2020 Apr; 21(4): 505–508. doi: 10.3348/kjr.2020.0146.
  9. Lippi G., Simundic A.M. European Federation for Clinical Chemistry and Laboratory Medicine (EFLM) Working Group for Preanalytical Phase (WG-PRE). The EFLM strategy for harmonization of the preanalytical phase. Clin Chem Lab Med 2018; 56:1660–6. https://doi.org/10.1515/cclm-2017-0277.
  1. Lippi G., Simundic A.M., Plebani M. Potential preanalytical and analytical vulnerabilities in the laboratory diagnosis of coronavirus disease 2019 (COVID-19). Clin Chem Lab Med. 2020 Mar 16. doi: 10.1515/cclm-2020-0285.
  2. Lippi G., von Meyer A., Cadamuro J., Simundic A.M. PREDICT: a checklist for preventing preanalytical diagnostic errors in clinical trials. Clin Chem Lab Med 2020;58:518–26. doi: https://doi.org/1 0.1101/2020.03.14.20036202.
  3. Paton R.S., Mahan Gi & Lourenco, Wikramaratna P. Estimating false-negative detection rate of SARS-CoV-2 by RT-PCR. medRxiv preprint. DOI: https://doi.org/10.1101/2020.04.05.20053355.
  4. Shen Z., Xiao Y., Kang L., Ma W., Shi L., Zhang L., et al. Genomic diversity of SARS-CoV-2 in Coronavirus Disease 2019 patients. Clin Infect Dis 2020. DOI: 10.1093/cid/ciaa203.
  5. Wang W., Xu Y., Gao R.; et al. JAMA. Published online March 11, 2020. doi:10.1001/jama.2020.3786.
  6. West C.P., Montori V.M., Sampathkumar P. COVID-19 Testing: The Threat of False-Negative Results. Mayo Clinic Proceedings (pre-proof). S0025-6196(20)30365-7. https://doi.org/10.1016/j. mayocp.2020.04.004
  7. Wolfel R., Corman V.M., Guggemos W., et al. Virological assessment of hospitalized patients with COVID-2019. Nature. 2020. DOI: 10.1038/s41586-020-2196-x.
  8. Xie X, Zhong Z, Zhao W, Zheng C, Wang F, Liu J. Chest CT for typical 2019-nCoV pneumonia: relationship to negative RT-PCR testing. Radiology 2020:200343. DOI: 10.1148/radiol.2020200343.
  9. Yi H. 2019 novel coronavirus is undergoing active recombination. Clin Infect Dis 2020. DOI: 10.1093/cid/ciaa219.
  10. Zou L., Ruan F., Huang M., Liang L., Huang H., Hong Z., et al. SARS- CoV-2 Viral Load in Upper Respiratory Specimens of Infected Patients. New England Journal of Medicine. 2020. pp. 1177– 1179. DOI: 10.1056/nejmc2001737.