False positives in SARS-CoV-2 disease

Dear Editor:

Since the description in January 2020 of the new coronavirus SARS-CoV-2, the detection of nucleic acids, by means of the real-time polymerase chain reaction (RT-PCR), has emerged as the infection’s first-line diagnostic tool.¹ Obtaining respiratory samples from deep nasopharyngeal smears, sputum and others are key elements in the definition of a case by this new coronavirus. In addition, the massive application of these tests has made it possible to trace microbiological dissemination routes in the description of clusters of cases, contributing significantly to containing the spread of this virus with its potential contagious power.

From the first papers, it was clear that the sensitivity of the SARS-CoV-2 infection detection test depended not only on the phase of the disease (detection being more likely in the first days after infection²) but also on correct sampling methods. In this sense, the performance of successive tests in patients with a clinical suspicion of COVID-19 allows the technique’s global sensitivity to improve, with a notable reduction in false negatives.³

Although the application of mass polymerase chain reaction (PCR) tests have proven to be an effective public health strategy for containing the pandemic, however, individual application to asymptomatic subjects can occasionally lead to misdiagnosis, and false positives. The hypothesis of a false positive result in an asymptomatic subject can be derived from statistical, laboratory and virological aspects, described below.

Regarding statistics, like the rest of the diagnoses the diagnosis of infectious diseases constitutes the establishment of a probability conditioned to the result of a test (Bayesian statistics). Regardless of the PCR’s format or commercial kit, it constitutes a test of very high specificity, and if we assume it is almost perfect (99.9%) it would lead to the practical absence of false positives. However, sensitivity and specificity refer to the characteristics of the test itself, but as doctors, we are more interested in the interpretation of the results, that is, establishing positive and negative predictive values (PPV and NPV respectively). As the assumed specificity and sensitivity of the PCR test for the diagnosis of COVID-19 are 99.9% and 70% respectively, important differences exist in the PPV according to the different pre-test probability (PPT). Thus, applying a PCR test to an asymptomatic low-risk patient (for example, with a PPT of 5%) would have a positive predictive value of ‘just’ 77.8%, while with the same test applied to an asymptomatic risk contact (with a PPT of 25%) would have a PPV greater than 96.5%. Therefore the medical application of the PCR-based molecular diagnostic tests have been mainly as a confirmatory diagnosis of patients with a high PPT. The case of SARS-CoV-2 infection constitutes the first global experience of the implementation of molecular tests in the epidemiological control of an infection, and although the result can be considered as valid in the public health system, the results for individual cases should be interpreted with caution by the doctor.

Diagnostic equipment can be another source of errors. The global urgency arising from the COVID-19 pandemic has resulted in different government agencies, such as the Food and Drug Administration (FDA) and the European Medicine Agency (EMA), to grant approval via the emergency route of numerous commercial kits. This has meant that evaluations in the clinical samples have been less exhaustive than usual, for which some authors have argued that the precision in real clinical practice of some of these kits might not be well established.⁴

Additionally, the PCR allows the detection of minute amounts of viral RNA, which can survive under favourable biological conditions for weeks and not necessarily correlate with active infection or contagiousness. In the case of the diagnosis of COVID-19, in some diagnositic tests, this is further aggravated by the use of genes common to other seasonal coronaviruses, whose positivity could be erroneously interpreted. In addition, there are different aspects of the behaviour of the virus that are still unknown and which may also condition the diagnosis. Since the start of the second wave, there has been considerable controversy about the significance of detecting small amounts of any of the genes, especially the N gene, in isolation.

The molecular microbiology labs have very strict protocols and work procedures so as to avoid contamination by nucleic acids and amplified acids, etc., which would condition the appearance of false positives. The automation of the process through robotic systems and the performance of internal laboratory controls further contribute to reducing the risk of contamination. However, these in-process controls can be affected by the very high workload, which in many cases requires the processing of thousands of samples per week. They may also succumb to the pressures derived from knowing that critical decisions in relation to economic, educational and social activities of a community depend on the reports issued by the microbiology lab.

Finally, there are also virological aspects to consider. An infection is the interaction between a microorganism and its host capable of inducing a pathogenic effect. The infective capacity of a microorganism not only depends on the virulence of the microorganism (amount of inoculum capable of causing an infection), but also on factors dependent on innate and acquired immunity. Thus, the detection of COVID-19 RNA in symptomatic patients is intrinsically associated with viral replication (active virions). However, this may be different in asymptomatic subjects.

SARS-CoV-2 virus RNA contaminates inert surfaces, sewage,⁵ animals including pets, etc. In addition, it has been detected in faeces, breast milk and others, but not all of them have been shown to correspond to sites of active viral replication. Several recent papers have shown a correlation between the existence of a high viral load determined by quantitative RT-PCR and positivity in viral culture. Also, a lower percentage of positivity has been shown in viral cultures from asymptomatic subjects than from symptomatic subjects, although the differences did not reach a level of significance. However, the usefulness of quantification also has its dark side, in that the viral load obtained is decisively conditioned by the quality of an inherently heterogeneous sample, which can directly vitiate the comparability of the results.

As doctors, we must learn to interpret the positive results of PCR for SARS-CoV-2 in asymptomatic subjects, depending on whether the patient is in the epidemiological context of risk contact, in which case we must implement isolation and clinical control measures, or is an asymptomatic subject with an isolated positive result. In the latter case, good communication with the microbiology services is essential as this will allow us to contemplate the possibility of a false positive, depending on the patient’s clinical characteristics and the specific microbiological parameters, and to establish strategies such as repeating the PCR, performing a PCR that acts on different targets, or obtaining non-molecular test results, maybe using an immunodiagnostic test. The implementation of these techniques can often avoid delays in the application of surgical interventions or on non-delayable invasive procedures, which can undoubtedly take a severe toll on a non-infected patient.