False-negative result in molecular diagnosis of SARS-CoV-2 in samples with amplification inhibitors

Introduction: Although reverse transcription-polymerase chain reaction (rRT-PCR) is the gold standard method for detecting severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), some factors, such as the presence of amplification inhibitors, lead to false-negative results. Objective: Here we describe the differences between rRT-PCR results for SARS-CoV-2 infection in normal and diluted samples, simulating the need for dilution due to the presence of amplification inhibitors. Material and method: Viral ribonucleic acid (RNA) from samples of nasopharyngeal swabs from 20 patients previously detected as “Negative” and 21 patients detected as “Positive” for SARS-CoV-2 was performed with the EasyExtract DNA-RNA kit (Interprise ® ). The rRT-PCR was performed with the OneStep/COVID-19 kit (IBMP), with normal and diluted (80 µl of H 2 O RNAse free) samples, totaling 82 tests. Results: The results indicate that there is an average variation ( α < 0.05) delaying the Cq between the results of amplification of the internal control (IC), N gene ( N G), and ORF1ab (OF), 1.811 Cq, 3.840 Cq, and 3.842 Cq, respectively. Discussion: The extraction kit does not completely purify the inhibitor compounds; therefore, no amplified product result may occur. In this study, we obtained a 19.04% false-negative diagnosis after sample dilution; this process reduces the efficiency of rRT-PCR to 29.8% in detecting SARS-CoV-2. Conclusion: Knowing the rRT-PCR standards of diluted samples can assist in the identification of false-negative cases and, consequently, avoid incorrect diagnosis. eficiência da rRT-PCR 29,8% na detecção SARS-CoV-2. Conhecer padrões da rRT-PCR de amostras diluídas auxiliar na identificação de casos falso negativos e, consequentemente, diagnóstico


INTRODUCTION
The first confirmed case of Coronavirus disease 2019  in Latin America occurred in Brazil, on February 25, 2020 (1) . Since then, until August 2020, Brazil has recorded about 4.1 million cases and about 126 thousand deaths due to COVID-19 (2) .
Early detection of infected individuals with large-scale testing, immediate isolation of screened cases, preventive self-isolation of close contacts, and prompt treatment for severe cases are essential measures to reduce the spread of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) (3) . Therefore, to quickly diagnose infections and mitigate the transmission of SARS-CoV-2, the real-time reverse transcriptionpolymerase chain reaction (rRT-PCR) is being used as the primary method in research and hospital laboratories to identify the virus in respiratory samples such as sputum or nasal, throat, nasopharyngeal swabs (4) .
The rRT-PCR tests typically take 4 to 6 hours to complete, with extraction, amplification, and detection of ribonucleic acid (RNA) (5) . Considering the limited supply of extraction reagents and test kits worldwide, extraction kits without RNA purification aim to solve this limitation and shorten the extraction time, thereby shortening the response time (4,5) .
Extraction kits without RNA purification may need to optimize rRT-PCR by sample dilution in case of problems with the rRT-PCR amplification, thus minimizing the presence of amplification inhibitors (7) , allowing amplification even in the presence of inhibitors or some sample degradation, avoiding the need to request a new sample from the patient, however, it is necessary to know the diluted amplification patterns, avoiding false-negative diagnosis.
Due to the severity of the pandemic, test kits were and are being developed and approved quickly to meet the worldwide demand for large-scale tests, generating the need for information on real data on the use of these kits in diagnostic laboratories (8) . Here we describe the differences between the rRT-PCR results for SARS-CoV-2 infection in normal and diluted samples, simulating the need for dilution due to the presence of amplification inhibitors.

Nasopharyngeal swab samples of RNA extraction
Samples of nasopharyngeal swabs from 41 patients admitted to the Ministro Costa Cavalcanti Hospital in Foz do Iguaçu, Paraná state, Brazil, were selected. Twenty of these patients were previously detected as "negative" and 21 patients were detected as "positive" for SARS-CoV-2 infection by the rRT-PCR diagnosis. The swabs were stored in tubes with 1× phosphate-buffered saline (PBS 1×), at -20ºC, until extraction.
The rRT-PCR assay was performed using the Biomol OneStep/ COVID-19™ Kit (IBMP), lot 200399Z074, 15 μl of reaction rRT-PCR Mix and 5 μl of purified sample RNA (from RNA extraction) or purified negative control, were pipette up and down to mix and, for positive control, 15 μl of the Mix reaction was mixed with 5 μl of the positive control (9) .
The results were evaluated by the rRT-PCR amplification standards, amplification values and submitted to descriptive analysis, normality test, and analysis of variance (Anova), to detect differences between the results before and after dilution.

The efficiency of the rRT-PCR
The analytical efficiencies for detecting SARS-CoV-2 from the normal methodology and after dilution were performed by serial dilution in the following proportions: 1, 1:2, 1:4, 1:8, and 1:10 (view Efficiency Curves in supplementary files). The results were evaluated by scatter plots and the efficiency values calculated from the R² of the linear regression (view Efficiency Test in supplementary files).

RESULTS
The results are described in Before dilution, samples 25, 26, 39, and 40 showed values lower than the cut-off stipulated for ORF1ab, and were considered positive. After dilution, they all shifted the ORF1ab Cq values to the right and were then considered negative due to non-amplification within the cut-off parameters ( Figure 1).

DISCUSSION
The positive control showed amplification for the three targets evaluated in all tests (Cq ≤ 35) and the negative control did not show any amplification for the three evaluated targets, according to the Mix manufacturer's protocol, validating the results.
The tests performed to demonstrate 1:2 dilutions were interesting to obtain a reliable amplification in samples with inhibitors, as shown in Figure 2. However, it is important to perceiving the result curve patterns after dilution.
In the example shown in Figure 2A, the sample without dilution did not achieve amplification of the IC (Cq = undetermined value) and OF patterns (Cq = undetermined value), and NG (Cq = 29.995) not defined and not showing a perfect exponential In cases as in Figure 2, only the dilution is sufficient to diagnose the sample as positive, avoiding new stress for the patient in repeating the collection, and new exposure by the health professional, the infected patient will be referred to the correct False-negative result in molecular diagnosis of SARS-CoV-2 in samples with amplification inhibitors  Efficiency curves N gene IC ORF1ab treatment site in a short period of time between collection and diagnosis. However, in cases where the result after the dilution is negative, a series of precautions should be taken when releasing the diagnosis, such as the characteristics of the curve, the graph must be evaluated completely and not only the values that exceeded the Cq.
According to the Thermo Fisher ® manual, considering a process efficiency of 100%, there is a known variation in Cq when the sample is diluted, which is variable according to the proportion of the dilution. This dilution variation can be ∆Cq = 1 from 1:2, ∆Cq = 2 from 1:4, ∆Cq = 3 to 1:8 and ∆Cq = 3.3 from 1:10. However, these values vary according to the efficiency of the process and presence of inhibitors (10) , which can result in false-negative diagnosis in low viral load samples, depending on the value used as a parameter to distinguish between positive and negative.
The positive samples tested that showed divergent results after dilution obtained ∆Cq values between 28 and 33, which should not make them negative after dilution even with low efficiency in the amplification process, since there was a variation between 8 and 12 Cq (Figure 1). Considering the progression of ∆Cq according to larger dilutions, the dilution proposed by the manufacturer of the viral RNA extraction kit (1:10) would not be interesting, as, theoretically, 1:10 would cause the Cq values to be even later. Larger tests involving smaller dilutions can be performed verifying in what proportion there would be no significant differences in Cq values and the effectiveness in the dilution of the rRT-PCR inhibitors.
The importance of performing rRT-PCR in kits that provide IC marking has already been reported by Kim et al. (2016) (10) , generating conclusive results about the extraction process, avoiding the release of false-negative results in samples that were not amplified with precision, since the interpretation of the results is not always direct. The sensitivity of rRT-PCR is negatively impacted by compounds present in the clinical sample that may partially or completely inhibit rRT-PCR chemistry (11)(12)(13)(14)(15) .
Protocols with purification steps can avoid the presence of amplification inhibitors, removing potential endogenous rRT-PCR inhibitors such as detergents, chelating compounds, and guanidine-HCl (11,13,(16)(17)(18)(19) . The efficiency of removing inhibitors from patient samples may be related to the intrinsic properties of the method used to extract the RNA (20) , which is not the case of the kit used in this study. The Easy Extract™ kit does not completely purify the inhibitor compounds, which significantly reduces the extraction time; however, non-amplification by inhibitors may occur.
A diagnostic error can lead infected patients to non-COVID-19 areas with the subsequent risk of infection for others areas; or patients who are SARS-CoV-2 negative to be sent to COVID-19 areas (21) , generating possible contamination to uninfected patients and also the spread of viruses in the disinfected areas, which can lead to viral spread within hospitals and treatment centers and contaminate health professionals. Knowing the rRT-PCR standards of diluted samples can help in the identification of false-negative cases and, consequently, avoid a wrong diagnosis.

CONCLUSION
The 1:2 dilution of the sample with inhibitors using the UltraPure ® H 2 O RNAse free generated amplification in 100% of the tested cases, therefore it is an alternative to avoid new sample collection from the patient. However, we emphasize that in this study we obtained 19.04% false-negative diagnosis after sample dilution, and this process reduces the efficiency of rRT-PCR to 29.8% in detecting SARS-CoV-2. It is possible to infer that the dilution helps in cases which a new sample collection is not feasible, but caution is needed in the evaluation of the rRT-PCR result.
It is important to assess the pattern of the amplification curves after dilution to avoid inaccurate diagnosis. If the sample with inhibitors is positive with a high viral load, the result will be reliable if IC and NG amplification occur up to Cq 30 and ORF1ab up to Cq 35. In case of non-amplification of the NG and ORF1ab curve after dilution, we recommend evaluate the need for a new sample and new analysis.