A Rapid RT-LAMP Assay for SARS-CoV-2 with Colorimetric Detection Assisted by a Mobile Application

Abstract

Loop-mediated amplification has been promoted for SARS-CoV-2 screening, however, antigen tests are preferred in low-income countries and remote zones. Poor training in molecular biology, plus the need for RNA purification or reading instruments to overcome issues of sensitivity in colorimetric detection, are some of the reasons limiting the use of this technique. In this study, nasopharyngeal swabs, aspirates and saliva were amplified in an in-house LAMP assay and subject to colorimetric detection, achieved by the naked eye and by image analysis with a mobile application. Accuracy of detection by the naked eye ranged from 61-74% but improved to 75-86% when using the application. Sensitivity of the digital approach was 81% and specificity 83%, with poor positive predictive value, and acceptable negative predictive value. Additionally to the reported effect of some transport media's pH, the presence of mucus and warming up of reagents while setting up the reaction critically affected performance. Accuracy per type of sample was 55, 70 and 80%, for swabs, aspirates and saliva, respectively, suggesting potential to improve the test in saliva. This assay, carried out in a closed tube, reduces contamination, has few pipetting steps and requires minimal equipment. Strategies to improve performance and implications of the use this sort of colorimetric LAMP for massive testing are discussed.

Keywords: COVID-19; SARS-CoV-2; cell phone application; colorimetry; isothermal amplification.

Figure 1

Agreement of sample pre-processing strategies. Percentages in each strategy are values of total agreement, calculated based on results for eight NPSs and four NPAs with Ct from 11–33 plus two negatives of each type of sample (labeled un); for esthetic reasons, only half of the pictures are shown. For all gels: smears were counted as positives, M = 100 pb ladder, numbers 11, 18, 23, 26, 30, 33 = Ct of the sample. For tubes for visual detection: “orangish” was taken as negative, numbers 11, 18, 23, 26, 30, 33 = Ct of the sample. For the application (App): order is for tubes shown for visual detection (Ct 11, 18, 23, 26, 30, 33, un, un), Fc (fold change) = after/before; if Fc > 1.0, the sample was classified as positive.

Figure 2

Effect of sample quality on visual detection. Categories correspond to: tp = true positives, tn = true negatives, fp = false positives, qRT-PCR negatives that turned yellow in LAMP, fn = false negatives, qRT-PCR positives that gave faint or negative visual outcomes. Percentage of agreement with qRT-PCR is 57.1, 62.4, 75.0, 50.0 and 62.0, for bloody, clear, fuchsia, mucus and precipitated conditions, respectively.

Figure 3

Relation between fold change and viral load. Y-axis indicates fold change values for a total of 208 samples, with the horizontal line corresponding to the threshold established. Boxes contain average and standard deviation of the fold change per Ct group, where Neg corresponds to the group with undetectable RNA of SARS-CoV-2 by qRT-PCR. The red line marks a fold change = 1.0.

Figure 4

Performance of the application-mediated detection among symptomatic and asymptomatic patients. Accuracy = tp + tn/total of symptomatic or asymptomatic in each Ct group; standard deviation was established based in the calculated accuracy per type of sample in each group. The red line marks a 85% of accuracy, required in Colombia for SARS-CoV-2 diagnostic tests.