Clinical validation of optimised RT-LAMP for the diagnosis of SARS-CoV-2 infection

Abstract

We have optimised a reverse transcription-loop-mediated isothermal amplification (RT-LAMP) assay for the detection of SARS-CoV-2 from extracted RNA for clinical application. We improved the stability and reliability of the RT-LAMP assay by the addition of a temperature-dependent switch oligonucleotide to reduce self- or off-target amplification. We then developed freeze-dried master mix for single step RT-LAMP reaction, simplifying the operation for end users and improving long-term storage and transportation. The assay can detect as low as 13 copies of SARS-CoV2 RNA per reaction (25-μL). Cross reactivity with other human coronaviruses was not observed. We have applied the new RT-LAMP assay for testing clinical extracted RNA samples extracted from swabs of 72 patients in the UK and 126 samples from Greece and demonstrated the overall sensitivity of 90.2% (95% CI 83.8-94.7%) and specificity of 92.4% (95% CI 83.2-97.5%). Among 115 positive samples which Ct values were less than 34, the RT-LAMP assay was able to detect 110 of them with 95.6% sensitivity. The specificity was 100% when RNA elution used RNase-free water. The outcome of RT-LAMP can be reported by both colorimetric detection and quantifiable fluorescent reading. Objective measures with a digitized reading data flow would allow for the sharing of results for local or national surveillance.

Figure 1

Stability test of improved RT-LAMP reaction mix. (a) Stability test of RT-LAMP reaction mix containing O117_N and O117_Q. Tubes were added with Synthetic RNA Control 2—MN908947.3 (‘+’) or human genome cDNA (‘−’) followed by heating at 65 °C for the time indicated by T. Each case has three replicates. (b) Dried reaction mixes stored at room temperature up to 3 days. Then they mixed with 20 µL DNase/RNase free water and 5 µL of Synthetic RNA Control 2—MN908947.3 (‘+’) or human genome cDNA (‘−’). The tubes were incubated at 65 °C for 30 min. Top row: Dried reaction mixes containing O117_N; Bottom row: Dried reaction mixes containing O117_Q.

Figure 2

Sensitivity and specificity of improved RT-LAMP assay. (a) Determination of 50% endpoints for O117_Q and O117_N preparations of RT-LAMP assay. Full length transcripts were serially diluted in buffer AVE to achieve the total RNA input/reaction indicated on the y axis. Each dot represents one experimental replicate. Dashed lines indicate 50% endpoint as calculated by the Reed-Muench method. Kits were dried except where indicated for RNA Control 2. Full length transcript indicated on figure. Dried reaction mix containing (b) O117_N and (c) O117_Q were tested against non-SARS human-infective coronaviruses. A and B represent two technical replicates.

Figure 3

Quantitative evaluation of pH-dependent colorimetric RT-LAMP readout. Three quantitative methods (a) Absorbance using 430/560 nm ratio; (b) SYTO 9 fluorescence using 485 nm (excitation) and 500 nm (emission); and (c) Qubit fluorescence using Qubit 2.0 fluorometer were used to assess the RT-LAMP result. (d–f) are the correlation analysis between RT-qPCR and the three quantitative evaluations, respectively. Each dot represents one experimental replicate. Dotted lines indicate 3× standard deviations above the negative controls.

Figure 4

Detection of SARS-CoV-2 from RNA extracts from swab and saliva specimen. Comparison of RT-LAMP for detecting SARS-CoV-2 RNA in extract vs RT-qPCR Ct value. RT-LAMP results were read by colorimetric changes (see Supplementary Fig. S5).