Introduction
The Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV2) is responsible for the coronavirus disease (COVID-19). Rapidly after its emergence at the end of 2019 in Wuhan, China, COVID-19 has being spread globally and becoming a massive pandemic [
1,
2]. COVID-19 pandemic has provided a great recognition to the key role played by the laboratories in the diagnostic testing; for screening, monitoring and contact-tracing [
3]. Indeed, molecular testing is the angular stone to prevent and control virus circulation: it is reliable, rapid and accurate. Through this pandemic, millions of RT-PCR tests were carried out daily, which demonstrate how suitable it is for large scale testing. As of 29 September 2021, 244.69 per 1000 inhabitants were tested by RT-PCR in Tunisia (total of 2.92 millions) which is a high rate as compared to other countries [
4]. Soon after the emergence of the SARS-CoV2, the first full sequence was published on January 10th, 2020 and the first RT-PCR detection assay was published on January 23rd, 2020 [
5]. Today, whole panoply of protocols and kits is available within reach of all laboratories; many other protocols have been published by research groups, number of them has been approved by the WHO in March 2020 and a huge number of commercial tests have been created by companies [
6,
7]. These tests are diverse by targeting a range of SARS-CoV2 specific genome regions. Obviously, genome targets impact the analytical performances; the sensibility of detection by RNA copy per milliliter and the specificity to detect unambiguously the SARS-CoV2 [
3]. The biggest challenge for laboratories was to have a sufficiently sensitive RT-PCR assay to be able to detect the virus in pre-symptomatic individuals, who harbor often low viral loads [
8]. Oppositely, high sensitivity may lead to false positive results especially for patients having recovered from SARS-CoV2 infection, due to genomic debris but no viable virus [
9]. Many authors focused on the evaluation and the improvement of available RT-PCR protocols and tests, but especially for commercial tests.
In this study, we aimed to evaluate and to compare 4 widely used WHO approved RT-PCR protocols on real clinical specimen, to decrypt the origins behind the diverging results and to propose recommendations for the choice of genome targets. We focused on the impact of the SARS-CoV2 Alpha variant of cocern (VOC) detection by the Institut Pasteur Paris protocol and how to manage it.
Discussion
In the present study, we provided a concise description of discordant results between 4 WHO approved RT-PCR protocols to detect the SARS-CoV2 genome: BERLIN, HKU, DAAN gene and IPP, through testing 260 real clinical specimens. The DAAN gene® commercial test used the same CDC China protocol approved by the WHO. In vitro analysis showed discordant results in 29.2% of cases (76 out of 260). The most discordant protocol is DAAN Gene® due to false positive late signals with N target. Discordant results between the two protocol’s targets are more frequent when viral load are low (high Ct values). Our results demonstrated that the multiplexing has worsened the sensitivity of the IP4 target. We provide concise recommendations for the choice of the targets, the interpretation of the results and the alarm signals which makes suspect a gene mutation.
Globally, molecular testing by PCR has revolutionized the diagnosis of infectious diseases. In the context of the COVID-19 pandemic, high performing tests has allowed it to identify infected people and to decide of their discharge. Different targets were proposed and validated, classically; the preferred targets of pathogens include the conserved specific genes as the nonstructural genes like RdRp or genes that are expressed abundantly such as the structural S and N genes. In our study, the protocols were approved since March 2020 by the WHO. Many authors studied these protocols to focus on sensibility and specificity by in vitro or in silico studies [
10‐
12]. Our study explored in vitro clinical specimen. Importantly, our study used a large number of clinical specimens from patients with confirmed COVID-19, oppositely, the majority of published studies explore cell culture supernatant or RNA transcripts, or sometimes a very small number of real samples collected in a precise time [
13].
In our study we found that 29.2% of tested samples gave discordant results (76 out of 260), occurring more often with high Ct values and probably due to false positive amplifications. Sule et al
. suggested that the Ct values over than 28 are probably related to non-specifically precipitated sequences due to an inactivation of the Taq polymerase and proposed that Ct values > 33.33 or 35, or ≥ 39.2 or 40 could be considered as negative [
14]. In our study, Ct value less than 40 is considered positive, according to the respective protocol authors.
Among all discordant cases, false negative results represent the most disturbing cases, for our study we obtained 21 samples positive by 3 protocols but one (Group D1 Table
2). This is problematic because not only it may underestimate the COVID-19 incidence but, perhaps more acutely, will lead to infectious individuals remaining as a source of infection in the community and undermine the effectiveness of infection control measures. Clinicians should not hesitate to re-sample the high suspected patients when laboratories return negative result.
The N gene target was the most problematic. From the one hand, this is due to the false negative results caused by mismatches with the primers and probes [
15]. Thanks to many in vitro and in silico analysis, researchers around the world are tracing the ongoing evolution of the N gene and demonstrated that it was particularly prone to mutations, more than all other targets. These findings affected the HKU, CDC China and Japan NIID N targets [
11,
15‐
17]. Wang et al. concluded in 14 september 2020 that the N gene is the most non conservative gene giving non uniform performances between different primers and probes, thus, the N gene may not be an optimal choice [
11]. This has become more preoccupant after the emergence of the Alpha VOC that caused a N gene dropout and N gene Ct value shift comparing to the wild strain [
18]. For the other hand, N gene target was widely criticized to give persistent late signals for convalescent patients, which make confusion between false positivity or low amount of virus, related to convalescent patients [
19,
20]. More importantly, this caused a dilemma regarding the discharge of isolation policies making the WHO canceling the need to 2 negative RT-PCR results [
21]. Moreover, soon after the publication of the United States of America CDC protocol, Lee et al. declared that the N gene target should be canceled and disused because of false positive reactivity of N3 [
22] This was found in our study for the DAAN gene® N target with late Ct values, which underlines that late positive signal for N gene alone is more likely to be false positive and needs to be interpreted with caution. In our series, the HKU N target didn't show so many false positive results; this would be related to the primers and probes design as they do not target the same region in the N gene.
Since its publication in January 2020, the BERLIN protocol was widely used worldwide [
5]. It was reported that the E gene was the most sensitive and the most conservative target [
11,
23] apart from some mutations affecting its sensitivity leading to false negative results by a commercial test using this protocol [
10].
BERLIN protocol was criticized for the low sensitivity of the RdRp target which was proposed to confirm all BERLIN E gene positive results [
24,
25]. Our results demonstrated that, in fact, 15 positive samples by BERLIN E target could not be confirmed by BERLIN RdRp target, in addition, BERLIN E gene gaves significantly lower Ct values than BERLIN RdRp leading to negative results for BERLIN RdRp when BERLIN E gaves late Ct values.
It is obvious that multiple RT-PCR reactions require more reagents, controls, thermo cyclers and labor, not adapted to a pandemic context. Multiplexing offers the possibility of two or more target detection by just one reaction, which have become attractive thanks to reducing significantly reagent consumption and time. Many authors suggested and demonstrated many successful multiplexing assays, without substantially reduce the test performances [
26‐
28]. However, in our study we demonstrated for the first time that multiplexing has drastically decreased the sensitivity of the IP4 target for particular samples. On 18 March 2021, the Centre National de Référence Virus des Infections Respiratoires (CNR), the authors of the IPP protocol, has reported a loss in the sensitivity of the IP4 target due to a mutation in the Orf1ab(C14050T), associated to the Alpha VOC [
29]. Our study samples were not sequenced, but in the routine surveillance of our lab, we noted that many samples with IP2+/IP4− belong to Alpha VOC, mainly detected by partial sequencing in the S gene [
30] and variant-specific RT-PCR tests SNPsig®real-time PCR SARS-CoV-2 mutation detection/allelic discrimination kit (Primerdesign Ltd; UK). Indeed, the samples with low Ct values with IPP IP2 target and negative amplification with the IPP IP4 target were collected between Week-7 and Week-16 of the year 2021; which corresponds to the high transmission period of the Alpha VOC in Tunisia (Chouikha et al. [
31]. The CNR IPP has published later in the 6 April 2021 a new RT-PCR Mix to overcome the false negative results. We demonstrated by experience that this abnormality was resolved by using the same targets in singleplex. This may be explained by a competition in the reaction which makes IP2 a preferred target and inhibits IP4 annealing. For this reason any multiplexing should be explored versus singleplex. Regarding the other targets, it was already reported that the Alpha VOC emergence has affected the sensitivity of some tests that amplify in the Spike gene, due to the deletion 69–70 [
32]. A large in silico study based on the theoretical production of RT-PCR signals by the SCREENED software, has evaluated the impact of the Alpha VOC on the RT-PCR protocols used in our study; the authors detected only one mutation (C12778T) in the Alpha VOC within the IPP IP2 target amplicon that has no impact on its sensitivity, as it is not located in the primers and probes annealing sites [
33].
Our study argued the need of using at least two independent virus key regions to avoid the false positive and negative results. The N gene is better to be avoided or should be interpreted with much caution. Otherwise, N late signals should be confirmed by testing another non structural region. The E target is the best one in terms of specificity; it presents the best agreement with non structural targets like IP2 and HKU Orf 1b with significant lower Ct values which may substitute the BERLIN RdRp to confirm positive E samples. Biologists should be aware about some alarm signals that should indicate close monitoring and investigating technical or molecular causes i.e. when using IPP or HKU protocol, any divergence between targets of more than 2 Ct values, when one target gives high positive signal while the other is totally negative and when a gene dropout is obtained for gene multiplex assays. Here, re-testing in singleplex should be performed as well as indicating sequencing. As the introduction of variants has worried the researchers about the reliability of RT-PCR protocols, which were established at the beginning of 2020 before the emergence of variants mutations, we recommend that all laboratories perform regular in silico analysis in order to assess the test performance. This could be limited for laboratories using commercial tests that do not specify exactly the primers and probes used.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.