Colorimetric RT-LAMP for SARS-CoV-2 detection from nasopharyngeal swabs or crude saliva: a multicountry diagnostic accuracy study in Africa

Abstract

Background: Reverse-transcription loop-mediated isothermal amplification (RT-LAMP) with colorimetric readout is a rapid, robust, and cost-effective one-step amplification assay that we previously trialled for the identification of SARS-CoV-2 in nasopharyngeal swabs in four countries. Here, we expanded our assessment of RT-LAMP for SARS-CoV-2 detection to several other African countries and evaluated its operational performance with crude saliva as a pragmatic approach for outbreak surveillance and response in resource-limited settings.

Methods: We conducted a multicountry diagnostic accuracy study of RT-LAMP for the detection of SARS-CoV-2 in different types of clinical samples. A preliminary study was conducted in Slovenia and Italy to establish the analytical performance (limit of detection) of RT-LAMP and optimise this assay before its deployment in Africa. Subsequently, we tested RT-LAMP with RNA extracted from nasopharyngeal swabs in seven countries in Africa (Angola, Burkina Faso, Côte d'Ivoire, Ethiopia, Senegal, Sudan, and Zimbabwe), and, in parallel, with crude saliva samples (ie, without RNA extraction) in an additional four countries (Cameroon, Ethiopia, Kenya, and Nigeria; paired nasopharyngeal swabs were collected at the same time). In both contexts, quantitative RT-PCR (RT-qPCR) with RNA extracted from nasopharyngeal swabs was used as the gold-standard benchmarking assay to evaluate performance. For RT-qPCR testing, each laboratory followed their own standard diagnostic procedure, whereas a standardised protocol was used for RT-LAMP. Saliva test standardisation was ensured through centralised reagent distribution. We calculated diagnostic parameters (sensitivity, specificity, and accuracy) using a 2 × 2 contingency table.

Findings: The preliminary study reported 87% sensitivity and 98% specificity for RT-LAMP. Between Sept 1, 2021, and June 30, 2022, we collected 2774 nasopharyngeal swabs and 577 crude saliva samples. For RNA extracted from nasopharyngeal swabs, the sensitivity and specificity of RT-LAMP for detection of SARS-CoV-2 (relative to the standard of diagnostics-ie, the RT-qPCR assay used in each participating laboratory) were 89% (95% CI 87-90) and 95% (93-96), respectively. Similarly, RT-LAMP tested on saliva without RNA extraction showed 80% (75-84) sensitivity and 99% (96-100) specificity (relative to the results obtained with the standard of diagnostics for RNA extracted from paired nasopharyngeal samples).

Interpretation: Colorimetric RT-LAMP is a reliable assay for SARS-CoV-2 detection in both extracted RNA and crude saliva samples. The demonstrably acceptable performance on crude saliva samples (without RNA extraction) underscores the scalability of this method for efficient outbreak surveillance in resource-limited settings.

Funding: Gates Foundation.

Figure 1

Study overview (A) Study scheme (left) and locations of participating countries conducting the two clinical trials (right). (B) Procedure for the clinical study on nasopharyngeal swabs; figure created with Biorender.com. (C) Procedure for the clinical study on saliva; figure created with Biorender.com. RT-LAMP=reverse transcription loop-mediated isothermal amplification. RT-qPCR=quantitative RT-PCR.

Figure 2

Viral loads in paired nasopharyngeal swab and saliva samples (A) Scatter plot shows low correlation between Ct values for paired nasopharyngeal swab and saliva samples from individuals diagnosed positive for SARS-CoV-2 as assessed by RT-qPCR (n=186). (B) Ct values from the plot in panel A were normalised for an internal control (UBC) in each sample (n=177; nine samples were excluded as no signal was obtained for UBC). (C) Box plot of Ct values from panel A. Boxes indicate limits of the first and third quartile; lines show the median; whiskers represent 1·5-times the IQR. (D) Ct values from panel C were normalised for an internal control (UBC) in each sample (n=177). Boxes indicate limits of the first and third quartile; lines show the median; whiskers represent 1·5-times the IQR. (E) Correlation of viral loads in positive saliva samples as assessed by RT-qPCR using two different assays in two laboratories at distinct timepoints (n=188; five samples were excluded from quantification as saliva volume was insufficient for additional analysis). Dashed lines in A, B, and E represent y=x. Ct=cycle threshold.RT-LAMP=reverse transcription loop-associated isothermal amplification.

Figure 3

Sensitivity of RT-LAMP for detection of SARS-CoV-2 in crude saliva samples (A) Sensitivity of RT-LAMP with crude saliva samples and RT-qPCR (LightMix) with RNA purified from saliva samples relative to the reference diagnostic method, stratified by RT-qPCR Ct values; n=144 for Ct <25; n=53 for Ct ≥25 (three positive samples [as assessed by RT-qPCR] were excluded due to an inconclusive colour change in RT-LAMP). (B) Sensitivity of RT-LAMP with crude saliva samples relative to RT-qPCR (LightMix) on RNA extracted from saliva, stratified by RT-qPCR Ct values. (C) Sensitivity of RT-LAMP as in panel A stratified by the number of genome copies per μL of saliva (upper plot) quantified with RT-qPCR (Luna); number of samples in each concentration range (count; lower plot). (D) Limit of detection of RT-LAMP for crude saliva samples lysed in saliva lysis buffer (n=193 SARS-CoV-2-positive individuals, as diagnosed by RT-qPCR on RNA extracted from saliva). Sensitivity values (thick lines) in A–C are shown with 95% CIs (whiskers). RT-LAMP=reverse transcription loop-mediated isothermal amplification. Ct=cycle threshold. RT-qPCR=quantitative RT-PCR.

Figure 4

Sensitivity and specificity of colorimetric RT-LAMP with RNA extracted from nasopharyngeal swabs RT-LAMP was conducted with RNA extracted from nasopharyngeal swabs (A) or with crude saliva samples (B). Sensitivity and specificity were calculated relative to the standard of diagnosis. Dashed lines represent the overall specificity or sensitivity of RT-LAMP relative to the standard of diagnosis. Error bars indicate 95% CIs. RT-LAMP=reverse transcription loop-mediated isothermal amplification.