Comparison of one-step RT-PCR systems
During the 2 weeks after the first PCR test on 27 January, a rapidly increasing number of case contacts in Bavaria was identified and their respiratory samples were tested. We found that the SARS-CoV E gene screening assay with the QuantiTect Virus +Rox Vial kit showed moderate to high amounts of unspecific signals in late cycles in 61% (451/743) of the tested patient samples and also of negative extraction and non-template controls (Table, Figure 2), which complicated the evaluation of the qPCR result. The RdRp assays were basically free from such unspecific signals in late cycles. Cycle threshold (Ct) values of the control SARS-CoV Frankfurt 1 RNA were reached three cycles earlier in E gene assay than in RdRp assays.
Table  Comparison of two different one-step real-time RT-PCR systems with SARS-CoV-2 assays from Corman et al. [5] and a commercial test kit with kit-specific assays, Bavaria, February 2020
Real-time RT-PCR system  PCR efficiency (%)a, linearity (R2)  Limit of detection (copies/reaction)  Unspecific signals count in E gene assay in totalb  Unspecific signals in E gene assay (%)b  Run time (hours)
QuantiTect Virus +Rox Vial kit (QIAGEN)  ND  ND  451/743(75/126 NC, 376/617 patient samples)  60.7  1:50
SuperScript III One-step RT-PCR System with Platinum TaqDNA Polymerase (Invitrogen)  95 / 0,99c  50c  13/257(2/38 NC,11/219 patient samples)  5.1  1:28
RealStar SARS-CoV-2 RT-PCR kit 1.0 (Altona)  125 / 0,97d  10d  0/111(0/38 NC, 0/73 patients samples)  0  2:15
NC: negative control samples; ND: not determined.
a E = 10−1/slope − 1.
b Indicated counts and percentage values of unspecific background signals in the SARS-CoV E gene assay are based on the total number of tested patient samples as well as the negative extraction and non-template controls.
c Only for RdRp gene assays, tested with four replicates of SARS-CoV Frankfurt 1 RNA [6]; 10-fold serial dilutions were determined. For the E gene, the assay was not linear.
d Only for the E gene, tested with two replicates of synthetic Wuhan coronavirus 2019 E gene control and SARS-CoV Frankfurt 1 RNA each [6]; 10-fold serial dilutions were determined.
Figure 2  Example image of real-time RT-PCR curves of the E gene assay with unspecific signals at late cycles, Bavaria, February 2020
RFU: relative fluorescence units.
Curves: 1: Wuhan coronavirus 2019 E gene positive control; 2: SARS-CoV Frankfurt 1 RNA positive control; 3,4,6,8: negative patient samples; 5: extraction negative control; 7: non-template negative control.
Signal is given in log scale with threshold = 200. PCR was performed with SuperScript III system and E gene primers and probe as published in [5]. Curves of positive controls (1 and 2) show expected sigmoid curves. Curves 3–6 show unspecific signals with increase above threshold. Curves below threshold were not considered as significant signals (7 and 8). Three additional one-step real-time PCR systems were compared with the initially used QuantiTect Virus +Rox Vial kit (QIAGEN) (Table): (i) OneStep RT-PCR kit (QIAGEN; data not shown in the Table because only a limited number of samples were tested in only one run), (ii) LightCycler Multiplex RNA Virus Master (Roche, Mannheim, Germany; data not shown because only a limited number of samples were tested in only one run) and (iii) SuperScript III One-Step RT-PCR system with Platinum TaqDNA Polymerase (Invitrogen, Darmstadt, Germany) (Table). Each assay protocol and thermoprofile was adjusted to the recommendations of the manufacturer, but primer and probe concentrations were the same as published [5]. They all showed comparable results in reduction of unspecific E gene signals. Using SuperScript III, the unspecific signals in the E gene screening assay were significantly reduced to 5% (13/257) in tested patient samples and negative extraction and non-template controls (Table), whereas Ct values for positive controls (Wuhan coronavirus 2019 E gene and SARS-CoV Frankfurt 1 genomic RNA [3]) were in the same range in QuantiTect (median: 29.1; range: 26.8–32.4; n = 85) and Superscript (median Ct: 28.1; range: 26.4–31.0; n = 30) setups.
In the following, we decided to switch to the SuperScript III system as recommended by Corman et al. [5], also because of its decreased thermoprofile run-time (Table). Moreover, we additionally included a newly launched commercial test kit in our study: RealStar SARS-CoV-2 RT-PCR kit 1.0 (Altona, Hamburg, Germany), which did not show unspecific E gene signals. A summary of the assay features of the three PCR setups tested and compared in detail is shown in the Table.
For evaluation of assay performance regarding efficiency, linearity and unspecific signals, we used the SuperScript protocol in a direct comparison of 73 samples with the RealStar SARS-CoV-2 assay. The E gene assays showed 97% (71/73) identical results in both assays (60 negative and 11 positive results as well as two divergent results, both with a positive Real Star SARS-CoV-2 assay and a negative SuperScript protocol assay). The specific SARS-CoV-2 assays gave in 67 (92%) identical results (60 negative and seven positive) and six divergent results. Overall, the RealStar SARS-CoV-2 kit seemed to be more sensitive.