Potential preanalytical and analytical vulnerabilities in the laboratory diagnosis of coronavirus disease 2019 (COVID-19)

Preanalytical errors

There is now incontrovertible evidence that the preanalytical phase is the major source of errors in laboratory testing, when used for either diagnostic [11] or research [12] purposes. Virology is not an exception, whereby many potential preanalytical errors are similar to those occurring in other different diagnostic areas, whilst others can be classified as specific. Among the former category, the safety and quality of RT-PCR testing may be endangered by lack of identification or misidentification of patient and/or sample, collection of inappropriate or inadequate material (for quality or volume), inaccurate conditions of sample transportation and storage (e.g. injury exposure, unreliable cold chain, prolonged transportation time), presence of interfering substances (e.g. release of cellular components that may interfere with the assay due to whole blood freezing, use of inappropriate additives) [13], [14], [15], as well as by a number of procedural issues occurring during sample preparation, thus including pipetting errors during manual sample preparation or aliquoting, cross-contamination and sample mismatch, among others [16]. The leading specific problems that may plague the quality of RT-PCR assays include sample contamination (even trace amounts of external DNA may jeopardize test results) and testing carried out in patients receiving antiretroviral therapy, which may then generate false-negative results [15].

Specimen collection and transportation

Among the various preanalytical issues that have been portrayed in the previous paragraph, those related to specimen collection are particularly significant and deserve specific focus. Although no detailed reference procedures have been provided as yet by the WHO for collecting respiratory material (i.e. lower, but especially upper respiratory specimens) for diagnosing SARS-CoV-2 to the best of our knowledge, the US Centers for Disease Control and Prevention (CDC) recommends that nasopharyngeal and oropharyngeal material shall be collected using swabs with a synthetic tip (e.g. as nylon or Dacron) and an aluminum or plastic shaft (other lower respiratory tract specimens could be collected, when available or feasible) [17]. The recommended procedure for collecting a quality nasopharyngeal specimen entails inserting the swab into the nostril parallel to the palate, maintaining the swab in place for few seconds for enabling secretion absorption and immediate placement of the swab into a sterile tube, containing 2–3 mL of viral transport media. The procedure for collecting oropharyngeal (e.g. throat) specimens entails swabbing the posterior pharynx, avoiding the tongue, and immediate placement of the swab into another separate sterile tube, also containing 2–3 mL of viral transport media. Failure to comply straightforwardly with the recommended procedures (e.g. use of wrong swabs, inappropriate absorption of diagnostic material, insertion into inadequate vials, contamination, and so forth) may be a significant cause of diagnostic errors, as clearly reported for other viral diseases [18], [19].

Diagnostic accuracy

Several assays have been developed so far for diagnosing SARS-CoV-2 infection. One of the most popular seem to be that originally proposed by the Charité-Universitätsmedizin Berlin Institute of Virology [20], and then endorsed by the WHO [21], along with that developed by the CDC [22], whose essential characteristics are summarized in Table 2, as reported in the respective websites [21], [22]. In the former case, the E gene assay is used as the first-line screening tool, then followed by confirmatory testing with an RNA-dependent RNA polymerase gene (RdRp) assay. The N gene assay can eventually be analyzed as an additional confirmatory assay. As regards the CDC test, the first panel, encompassing three N gene primer/probe sets, is designed for both universal detection of SARS-like coronaviruses (one primer/probe set), as well as for specific detection of SARS-CoV-2 (two primer/probe sets). An additional primer/probe set for detecting human RNase P gene (RP) in control samples and clinical specimens is included in the panel. Many other assays have then been developed by independent research institutes and in vitro diagnostic companies around the world, as summarized elsewhere [4].

According to recent evidence, the diagnostic accuracy of many of the currently available RT-PCR tests for detecting SARS-CoV-2 may be lower than optimal (i.e. <100%). Xie et al. first described the case of five out of 167 patients (3.0%) with chest computed tomography (CT) evidence of COVID-19, who initially tested negative for SARS-CoV-2 RT-PCR [23]. Interestingly, repeated swab tests carried out during hospitalization gradually turned to be positive in all such patients, with a mean interval period for positivity of 5.0±2.7 days. Ai et al. carried out another study, including 1014 suspected COVID-19 cases who underwent multiple RT-PCR testing and chest CT [24]. Overall, 88% (888/1014) of patients had positive chest CT scans, whilst RT-PCR positivity was found in 59% (601/1014) of all cases. As many as 34.7% of patients with positive chest CT findings had negative RT-PCR results of throat swab samples. According to clinical history and serial CT features, 11.6% and 16.6% of all patients with initially negative RT-PCR results were finally considered as probable or highly likely COVID-19 cases. Importantly, as many as 93% of all patients whose RT-PCR became positive for SARS-CoV-19 after an initially negative test result had CT features suggestive of COVID-19, with a mean interval period of 5.1±1.5 days for turning positive.

The fact that RT-PCR testing may be initially negative in patients with SARS-CoV-2 infection, especially in those who will later develop overt COVID-19, is not really surprising considering the probable kinetics of SARS-CoV-2 infection. Reliable evidence suggests that the incubation period of SARS-CoV-2 is around 6 days (interquartile range [IQR], 2–11 days) [25], and that the median period between symptom onset and hospital admission is 7 days (IQR, 4–8 days) [26], whilst the median period of symptom duration is around 13 days (IQR, 5–24 days), slightly longer in patients with severe disease (16 days; IQR, 10–20 days) [27]. Convincing information is also accumulating from China and abroad in support of the evidence that human-to-human contagion may be relatively rare, but not impossible, during the non-symptomatic phase of SARS-CoV-2 infection, whereby the virus could be occasionally transmitted during incubation by patients with brief and non-specific illness, including children, in whom the severity of COVID-19 is usually milder than in adults [28], [29], [30], [31]. This is supported by evidence that high viral loads, more in the nose than in the throat, can be detected soon after symptom onset, when the patient has not received a diagnosis of COVID-19 and has not been isolated, but can also be found in asymptomatic patients [32]. It is also worth mentioning here that virus shedding in some patients may continue for some days after symptom relief and recovery [33], [34]. Notably, a very recent study showed that the communicable period (expressed as first time of SARS-CoV-2 positive to date of virus clearance) in patients infected by SARS-CoV-2 was 6 days (IQR, 2–12 days) in subjects without symptoms compared to 12 days (IQR, 12–14) in those who became instead symptomatic [35].

Therefore, combining this epidemiologic evidence with the analytical sensitivity of the currently used RT-PCR assays, it is not surprising that at least two gray zones could be identified, potentially plagued by false SARS-CoV-2 negativity attributable to the low viral loads especially in asymptomatic or mildly symptomatic patients (Figure 1). The former would correspond to the initial phase of infection, when the patient is still completely asymptomatic or only mildly symptomatic. Virus shedding may have already initiated during this period, thought its extent would probably be too low to be identified by some RT-PCR assays (Figure 1). The second period would instead reflect the tail of SARS-CoV-2 infection, when there is symptom relief. In this final phase of the infection, virus shedding may still persist, though remaining below the analytical sensitivity of some RT-PCR assays (Figure 1).

Figure 1:

Correspondence between development of viral load during severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection, clinical course and positivity of (real time) reverse transcription polymerase chain reaction (rRT-PCR) assays.

Another important concern, which has recently been highlighted, is the risk of active recombination and mutations, which are attributable to the error-prone RNA-dependent RNA polymerases of coronaviruses. Shen et al. recently found a remarkable level of viral diversity in some infected patients, accounting for a median number of 4 intra-individual viral variants, which is suggestive of the rapid evolution of SARS-CoV-2 [36]. In another study, Yi detected as many as five different SARS-CoV-2 haplotypes, a fact that usually reflects active genetic recombination [37]. Such a viral evolution not only would explain the heterogeneity observed in intra-individual immune response, virulence, pathogenicity and transmissibility [38], [39], but the risk of mutation rate changes may also compromise the accuracy of RT-PCR detection.

Beside these microorganism-related issues, and like other areas of diagnostic testing [40], the accuracy of RT-PCR can be substantially plagued by lack of harmonization (of primers and probes) [41], as well as by a variety of technical and analytical errors, as summarized in detail elsewhere [13], [14], [15]. In brief, these typically encompass instrument malfunctioning (including inappropriate PCR cycling conditions), use of insufficient or inadequate material, non-specific annealing of PCR to homologous sequences, misinterpretation of expression profiles and so forth.