Performance of Loop-Mediated Isothermal Amplification (LAMP) Targeting the Nucleocapsid (N) Gene of SARS-CoV-2 for Rapid Diagnosis of COVID-19: Systematic Review and Meta-Analysis

Abstract

SARS-CoV-2 emerged as a new respiratory virus spreading rapidly to all areas of the world. A systematic review with meta-analysis concerning the use of Loop-Mediated Isothermal Amplification (LAMP) methodology targeting the SARS-CoV-2 nucleoprotein (N) gene was conducted. The search resulted in 229 articles, of which 19 articles were selected to compose the final review. In general, LAMP showed a high specificity in the detection of SARS-CoV-2 and a wide variation of sensitivity values. The LAMP method was considered a fast and highly specific method for SARS-CoV-2 detection; however, some variables may affect its sensitivity and overall performance.

1. Introduction

Starting in December 2019, the world went on alert with the announcement of a new coronavirus in the province of Wuhan, China. The new coronavirus spread rapidly, causing a variety of non-specific respiratory symptoms, in some cases progressing to fatal outcomes [1]. In 2020, the World Health Organization (WHO) declared the state of a world pandemic and several countries started research projects and investigations to better understand the spread and pathogenicity of the so-called SARS-CoV-2 [2]. To contain the spread of the virus, various protective measures have been applied, such as hand washing, wearing face masks, containment, and social distancing. However, there have been thousands of cases with the severe forms of COVID-19 and, as a result, a high number of deaths have been reported worldwide [1]. During the pandemic, several diagnostic strategies were proposed, mainly based on immunological and molecular biology methods [3,4].

The main methodology, considered the gold standard for SARS-CoV-2 detection, is Reverse Transcription, followed by real-time Polymerase Chain Reaction (RT-PCR). RT-PCR is considered extremely sensitive, allowing detection of the viral genome in respiratory samples during the early infectious phase of COVID-19 [5]. However, even RT-PCR is a time-consuming and expensive technique that requires dedicated equipment, which often makes it unfeasible for some diagnostic centers, especially in countries with limited resources [6].

Developed in 2000, loop-mediated isothermal amplification (LAMP) is a fast molecular method, with target sequence amplification in approximately 30 min. The technique requires only a simple heating block at a fixed temperature of around 65 °C for execution, maintaining a high sensitivity and specificity and at a lower cost compared to other molecular methods [7,8]. The addition of a reverse transcription enzyme makes the amplification of RNA possible by LAMP [6]. Since then, the technique has been used in the molecular identification of several pathogens, mainly in developing countries, affected by endemic tropical diseases [9,10,11,12]. LAMP reactions can also be read by different methods, including visual reading by color or turbidity changes, or using fluorophores to combine to double-stranded DNA fragments allowing visualization under UV light or in a dedicated instrument. This adds to the versatility of LAMP as compared to RT-PCR. Despite the advantages, LAMP methods may suffer from a lack of sufficient sensitivity, may be more prone to cross-contaminations and require more expertise to differentiate some positive and negative reactions, especially when color changes are not clear enough.

SARS-CoV-2 is a single-stranded RNA virus with an approximate length of 30 kb. The N gene, which encodes the nucleocapsid protein, was found at the 3’ end of the RNA strand, thus being less vulnerable to the action of RNAses that degrade the viral RNA from the 5’ to 3’ sense during the extraction and purification processes, thus ensuring viral amplification of SARS-CoV-2, even if it is partially damaged [6].

This study aimed to select and analyze published articles that used RT-LAMP methodology targeting the viral N gene, comparing its sensitivity and specificity through a systematic review with meta-analysis, to obtain insights into the performance of this methodology and the utility of the N gene in the detection of SARS-CoV-2 for clinical applications.

2. Materials and Methods

2.1. Article Search and Selection Strategy

For the systematic review and meta-analysis, searches were performed for articles in Medline (PubMed), Science Direct, LILACS and SciELO databases without language restrictions, published until 31 July 2021. The following keywords were used: [(SARS-CoV-2) OR (coronavirus infections)] AND [(Reverse Transcriptase Loop-Mediated Isothermal Amplification) OR (LAMP)]. For each search, the same keywords were used, with strategies compatible with the search tools of the respective databases, in addition to using full text filters in the search in Medline and open access in Science Direct. After the search, the articles found were added to the Rayyan program [13] to assist in article selection and identification of duplicate files.

Two researchers individually selected eligible articles to compose the analysis and a third researcher assisted with articles that received discordant opinions.

For the selection of articles, the researchers read the titles and abstracts separately. The following inclusion criteria were used for the selection of articles in this study: original (observational) diagnostic studies using the LAMP method targeting the N gene for the detection of SARS-CoV-2 in human samples from symptomatic and non-symptomatic patients and with demonstration of the percentage of sensitivity and specificity of the method or the number of true and false positive and negative cases found in comparison with the results obtained by RT-PCR.

Review articles were removed from the analysis, as well as articles that did not include the LAMP method, articles that addressed regions other than the N region of the virus, and articles that did not mention sensitivity and specificity, did not analyze clinical samples, or did not compare the performance of the technique LAMP with RT-PCR.

2.2. Data Analysis

A Microsoft Excel spreadsheet was used to assist in data analysis, in which the following information was extracted: sensitivity and specificity, target genes, type and number of samples, type of reading of the results, execution time and limits of detection (LOD).

The meta-analysis consisted of sensitivity and specificity data, and construction of the forest graph plot was performed using RevMan version 5.4 [14]. For this, the number of true positive and false negative cases, referring to the sensitivity and the number of true negative and false positive cases, referring to their specificity, were used for the construction of the graph.

3. Results

As a result of the initial search, we identified 229 articles in all searched databases. The search in Medline (PubMed) resulted in 196 articles, in Science Direct 25, in LILACS 4 and in SciELO 4. Of these, 22 articles were promptly excluded because they were duplicates. In the first selection round, which included reading the title and abstract, 165 articles were excluded for not meeting the pre-defined eligibility criteria. After this initial selection, 42 articles were considered eligible and read in full. Of these, 23 articles were still excluded because they did not present all the necessary information, defined in the present study, for the meta-analysis, and because they did not compare with RT-LAMP, did not use clinical samples, or did not target the N gene. Finally, the search resulted in 19 articles to compose this review (Figure 1).

A listing containing the 19 articles that make up the review is shown in

Table 1. Summary of data from articles that used the LAMP methodology to identify SARS-CoV-2 in symptomatic and asymptomatic patients included in this review.

Author	Target Gene	Nº of Samples	Type of Sample	Sample Pre-Treatment	Sensitivity % (CI:88. 0–92. 1)	Specificity % (CI:97. 9–99. 2)	Reading the Results
Alvarez, 2021 [15]	Orf1a, N, HMS Assay 1e	59 (29+/30−)	Swab NP and saliva (with or without purification)	RNA extraction and direct assay	93.1% (purified) 65.5% not purified	100.0%	Colorimetric
Baek, 2020 [16]	N	154 (14+/140−)	Swab NP, sputum and tears	RNA extraction	100.0%	98.7%	Colorimetric
Ganguli, 2021 [17]	Orf1a, S, Orf 8, N	20 (10+/10−)	Swab NP	Without RNA extraction	100.0%	100.0%	Fluorescence
Haq, 2021 [18]	Orf-1ab, N, S	297 (124+/173−)	Swab NP	RNA extraction	91.5%	90.0%	Colorimetric
Jang, 2021 [19]	RdRP, N, E	292 (130+/162−)	Swab NP, oropharyngeal smears, sputum, saliva	RNA extraction	93.9%	100.0%	Fluorescence
Jiang, 2020 [20]	N	260 (47+/213−)	Swab NP, sputum and tears	RNA extraction	91.4%	99.5%	Fluorescence
Kitagawa, 2020 [21]	N	76 (30+/46−)	Swab NP	RNA extraction	100.0%	97.6%	Turbidimetry
Kitajima, 2021 [22]	N	151 (79+/72−)	Swab NP and sputum	RNA extraction	88.6%	98.6%	Turbidimetry and Fluorescence
Lalli, 2021 [23]	N, E	30 (20+/10−)	Saliva	Pre-treated without RNA extraction	85.0%	90.0%	Colorimetric
Lee, 2020 [24]	N	157 (107+/50−)	Swab NP	RNA extraction	87.0%	100.0%	Fluorescence
Li, 2021 [25]	N	45 (15+/30−)	Swab NP	Without RNA extraction	100.0%	100.0%	Colorimetric
Manzano, 2021 [26]	N	183 (127+/56−)	Swab NP	RNA extraction	91.0%	100.0%	Fluorescence
Roumani, 2021 [27]	ORF8, N, ORF3a	152 (49+/103−)	Swab NP	RNA extraction	63.3%	99.0%	Fluorescence
Rödel, 2020 [28]	RdRP M, E, N	73 (38+/35−)	Respiratory Secretions	Without RNA extraction	81.6 %	100.0%	Fluorescence
Schellenberg, 2021 [29]	N	101 (50+/51−)	Swab NP	Without RNA extraction	77.0%	100.0%	Colorimetric
Dao Thi, 2020 [30]	N	235 (35+/200−)	Swab NP	Without RNA extraction	86.0%	99.5%	Colorimetric
Yang, 2021 [31]	N, AS1E	573 (278 +/295−)	Saliva	Pre-treated without RNA extraction	74.8%	100.0%	Colorimetric
Zhang 2021 [32]	ORF1ab, N	12 (8+/4−)	Swab NP	No RNA extraction (pre-treatment with NAOH solution)	100.0%	100.0%	Lateral Flow Assay (LFA)
Zhu, 2020 [33]	ORF1ab, N	129 (33 +/96−)	Oropharyngeal smear	RNA extraction	100.0%	100.0%	Nanoparticle-based lateral flow biosensor (LFB)

.

Nasopharyngeal swabs collected with swab (Swab NP); Lateral Flow Assay (LFA); Nanoparticle-based lateral flow biosensor (LFB); Confidence Intervals (CI).

Using the specificity and sensitivity data extracted from the selected articles, it was possible to build a forest graph plot using the RevMan 5.4 software [14]. The sensitivity and specificity of the different studies are shown in Figure 2.

4. Discussion

Although the LAMP test results can be evaluated in a variety of ways, for the purpose of this study, the performance of the LAMP methods described in the selected studies was evaluated in terms of sensitivity and specificity when using the N gene region as the molecular target, thus providing indication of the utility of this region for LAMP-based methodologies for the detection of SARS-CoV-2 from clinical samples.

In general, all retrieved articles showed high specificity for the detection of SARS-CoV-2 using RT-LAMP (97.6% to 100%). Analytical sensitivity, however, varied widely between studies (63% to 100%). Lower sensitivity values were related to RT-LAMP assays that used clinical samples directly, without prior RNA extraction, and to samples with very low SARS-CoV-2 viral load, usually with RT-PCR CT values above 35 [15,23,27,29,31]. Only one study using clinical samples directly, without prior RNA extraction, showed 100% RT-LAMP sensitivity compared to RT-PCR, but the total number of analyzed samples was only 45, which may have overestimated the overall performance of the test [25].

Most of the selected articles in this study showed promising results for the amplification of SARS-CoV-2 by RT-LAMP, with excellent specificity, as demonstrated in the forest graph plot (Figure 2). However, only Kitagawa et al. and Haq et al., out of the 19 selected articles, pointed to a more sensitive than specific RT-LAMP test [18,21].

The study by Kitagawa et al. had an excellent overall performance, with 100% and 97.6% sensitivity and specificity, respectively, in just 35 min of reaction time, along with an easy reading of the results by visual inspection of turbidity. However, the authors warned of a possible bias, related to variations in the concentration of RNA in different extracts, given the inconsistency found in two samples, whose results for RT-LAMP were positive, while the RT-PCR showed negative results [21].

The study by Haq et al. used 297 samples from Pakistani repatriated patients with suspected COVID-19, tested with both RT-LAMP and RT-PCR methods. The colorimetric RT-LAMP assay developed by the authors had a sensitivity of 91.45% and a specificity of 90% [18]. Reiterating the performance demonstrated by previous articles, it showed high sensitivity and specificity. This may be due to the concentration of RNA in the samples or their degradation.

In contrast, Roumani et al. showed excellent agreement in terms of specificity (99%), but a slightly lower sensitivity (63.3%), compared to RT-PCR. The authors argued that the false-negative results of RT-LAMP, with positive RT-PCR, were due to a low viral load among these samples, suggesting that these may represent potentially non-infectious samples, and thus the negative result of RT-LAMP should not affect the management of these patients [27]. Similar outcomes were reported by Schellenberg et al., Kitajima et al. and Thi et al., who demonstrated that the sensitivity of RT-LAMP tests compared to RT-PCR in clinical specimens is dependent on the samples viral load. In general, samples with a detection limit around a cutoff threshold (CT) of 30 to 36 cycles by RT-PCR correlated with a low viral load, generating false-negative results in RT-LAMP [22,29,30].

Baek et al. further evaluated the specificity of the primers, targeting the SARS-CoV-2 N region and used their RT-LAMP test against other coronaviruses such as SARS-CoV and MERS-CoV, in addition to other respiratory viruses. The authors reported no amplification in the RT-LAMP assay for any of the tested viruses; therefore, no cross-reactivity of the developed primers was observed with other respiratory viruses [16].

In the final selection of 19 articles, 4 resulted in “not estimable” according to the forest plot, as they presented 100% concordant RT-LAMP and RT-PCR results. Only a limited number of samples was used in these studies to compare with RT-PCR, which may have interfered with the calculation of the actual specificity and sensitivity values, whose results are questionable.

Considering the reaction time required for SARS-CoV-2 detection by RT-LAMP, we found a time variation ranging from 30 to 60 min, which, in any case, makes RT-LAMP an extremely fast method, even when compared with RT-PCR, which, in general, requires about 120 to 160 min to be completed, depending on the protocol being used [6,32].

The observed limit of detection (LOD) across the studies showed a wide range of values with the different LAMP assays, with LOD ranging from 2 to 500,000 copies of viral RNA per reaction [15,29]. This may be evidence of deficient standardization for some of these tests, since some studies used synthetic SARS-CoV-2 RNA to determine the LOD, whereas others did so using clinical samples with a viral load estimated by RT-PCR. In line with this observation, a similar result was reported from a meta-analysis report that compared the performance of the RT-LAMP methodology with digital PCR and quantitative PCR, where the authors found LAMP LOD values ranging from 80 to 10,000 copies per mL, indicating a rather lower analytical sensitivity for this methodology [34].

Two studies analyzed saliva as clinical samples for the LAMP colorimetric assay, pointing out a bias that could be a potential problem in the interpretation of LAMP results. According to the authors, naturally acidic saliva can lead to false-positive results with RT-LAMP due to changes in the pH indicator by an acidic ample, and not by the acidification of the reaction solution due to the amplification of the target sequence [15,31]. To correct the pH, sodium hydroxide (NaOH) was used by the authors, raising the pH of the buffer solution, thus correcting this problem, resulting in a test with 85% sensitivity and 100% specificity for the N gene, demonstrating that saliva can be a good alternative sample for detection of SARS-CoV-2 [31]. Using saliva, Lalli et al. observed that the 1:2 dilution of the sample in buffered saline, associated with pretreatment with RNase inactivation reagents, guanidine and thermal incubation at 65 °C for 15 min, potentially reduced the reaction inhibitors present in the saliva [23].

The development of new, fast, and low-cost molecular methods, while maintaining excellent sensitivity and specificity, is highly desirable, but rigorous standardization and investigation should be performed to identify the best gene targets, limit of detection, reproducibility, and potential application to different samples. One of the advantages highlighted for LAMP is the use of four to six primers, which recognize six to eight distinct targets of the SARS-CoV-2 viral RNA, making this method extremely specific.

5. Conclusions

Although it appears less sensitive than RT-qPCR, the RT-LAMP assay is a simpler option, which has been shown to be extremely specific for the amplification of SARS-CoV-2. Despite the reduction in the sensitivity of RT-LAMP in samples with a very low viral load, the method is a suitable tool for investigating SARS-CoV-2 infection in symptomatic patients. The utility of very low viral loads is debatable, with some authors arguing that individuals presenting with this situation may not transmit the virus; thus, a negative result would not affect the management of these patients, although it may represent an early infection stage, with patients becoming symptomatic thereafter. Nevertheless, another advantage of RT-LAMP demonstrated in some studies is the fact that it does not require prior nucleic acid extraction, reducing sample handling and processing costs. As it is an easy-to-perform molecular method, with the possibility of visual detection, RT-LAMP methodology can be an interesting alternative for laboratories with few resources and real time PCR equipment, or even, as a field methodology, in distant locations and large centers, providing fast results, which can be very useful for the management of infected patients.

In summary, we concluded that there is a need to include more samples in the validation of the tests in order to increase their reliability, in addition to new studies that seek to improve the sensitivity of RT-LAMP, using, for instance, combinations of primers for different target regions of the viral RNA or different clinical specimens. Constant improvements in the LAMP technique are indispensable for the consolidation of this test as a robust technique for the detection and identification of SARS-CoV-2 and other viruses of clinical interest.