Pooled analysis of LAMP assay for the diagnosis of norovirus infection

Abstract Background Rapid laboratory detection is essential to diagnose norovirus infection. LAMP has many advantages compared with RT‐PCR for detecting norovirus, including high sensitivity, high specificity, rapidity, low cost, and intuitive results, which can be easily read with the naked eye with the help of color‐based reporters. In this study, we intend to analyze the accuracy of LAMP methods for the diagnosis of norovirus infection. Methods Two researchers independently retrieved relevant literature up to January 2021 (PubMed, Web of Science, Cochrane Library, Embase, CNKI, Wan Fang, and VIP). The researchers screened all articles and extracted their research data for meta‐analysis. QUADAS‐2 tool was used to evaluate the quality of the included studies by Review Manager 5.3. Forest plots were performed by Meta‐DiSc 1.4 to evaluate the accuracy of the test. Deeks’ funnel plot symmetry tests were conducted by Stata 15.0 to check the potential publication bias. Results Eleven sets of data extracted from the eight included studies were included for meta‐analysis. For the detection of norovirus, the pooled sensitivity, specificity, positive LR, negative LR, diagnostic OR, and their 95% CI were 0.96 (0.95–0.97), 0.99 (0.99–1.00), 91.14 (31.88–260.56), 0.06 (0.04–0.09), and 1473.68 (562.96–3857.70), respectively. Besides, AUC in the SROC curve was 0.9920. Conclusion LAMP had high sensitivity and specificity in terms of the diagnosis of norovirus infection. However, further extension of this approach should be researched to ensure the accuracy and practicability of this hopeful test in the future.

leading 35, 000 deaths worldwide in 2010 and constituting the most deaths among the diarrheal diseases. 2,4,5 Norovirus was divided into at least 6 genogroups (GⅠ-GⅥ), and GⅠ and GⅡ were the predominant genogroups that caused acute gastroenteritis in humans. [6][7][8] Norovirus infection usually caused a self-limiting illness in healthy adults with 2-to 3-day course, while in the elderly, young children, and immunocompromised individuals, it was associated with severe complications such as developing severe disease and mortality. 6,7 Human norovirus has a substantially high prevalence and mortality across both healthcare and community settings, particularly in developing countries, and brings humans a severe global burden of disease. 5,9,10 Since the rapid spread of norovirus is a major public health issue, rapid laboratory diagnosis is essential to facilitate the execution of appropriate control measures to reduce transmission and outbreaks of the virus. The methods for the detection of norovirus can be concluded in terms of electron microscopy, immunology, and molecular biology. 8,11 Unfortunately, electron microscopy was not widely available for detecting norovirus in microbiology laboratories because of its low sensitivity, specificity, and expensiveness. 11 The development of enzyme immunoassays, such as IDEIA and RIDASCREEN, has been challenging because norovirus genotypes that formed most virus antigens were many (n = 29) and had the antigenic drift in certain strains. 8 In addition, due to the problem of sample processing, the IDEIA norovirus assay was at low sensitivity and raised serious questions regarding its usefulness in routine screening for norovirus. 12 Currently, reverse transcriptionpolymerase chain reaction (RT-PCR) is known as the gold standard in the detection of norovirus due to its high sensitivity and specificity in clinical and environmental samples. 6,8 However, the disadvantages of RT-PCR-based assays are also apparent, including high cost, time-consuming, and dependence on sophisticated equipment and expertise, leading to the limitation in its use in resource-poor settings without equipped laboratories. 11,13 Loop-mediated isothermal amplification (LAMP), created by Notomi et al., 14 was a remarkable alternative to RT-PCR for the diagnosis of norovirus infection. LAMP allowed RNA detection in an isothermal condition by using Bst DNA Polymerase that had high displacement activity, which was performed with simple and single-temperature incubation sources and thus got rid of the limitation of expensive equipment. 6,13,14 LAMP has many advantages compared with RT-PCR, including high sensitivity, high specificity, rapidity, low cost, and intuitive results, which can be easily read with the naked eye with the help of colorbased reporters. 15 While many LAMP assays have been presented to detect norovirus by research groups, it is essential to systematically evaluate and conclude the performance of LAMP and the quality of these studies. So far, no overall analysis of the accuracy of LAMP for norovirus seems to have been researched. Therefore, here, we intend to determine the accuracy of LAMP methods for the diagnosis of norovirus infection, using systematic review and meta-analysis techniques.

| Electronic searches
A systematic evaluation was conducted in this study according to the PRISMA guidelines. [16][17][18] Two researchers independently retrieved relevant literature up to January 2021 by using the databases, including PubMed, Web of Science, Cochrane Library, Embase, CNKI (China National Knowledge Infrastructure), Wan Fang, and VIP. The first four were in English, and the next three were in Chinese. The primary keywords combined LAMP with norovirus. The search had no restriction in language, but synonymous extensions were utilized.

| Study screening and selection
According to the predefined inclusion and exclusion criteria, two researchers (Xi-Feng Qian and Nan-Xi Li) as a group independently screened all English articles, including titles, abstracts, and full texts.
Another group (Ai-Ling Duan and Rong-Xian Huang) were responsible for all Chinese articles in the same ways. In addition, we scanned the bibliographies of most publications included to identify additional studies. All of them used Endnote X9 to manage and screen the literature. Then, they checked and rechecked the results. A third researcher would join to achieve consensus if there was still a discrepancy after discussing in each group.

| Data extraction
While reading the included articles independently, data were extracted simultaneously on author, year, study design, country, detection method, gold standard, sample source, sample type, sample size, incubation temperature (°C) and time (min), detection limit, genogroup type, true positive (TP), false positive (FP), false negative (FN), and true negative (TN). The third researcher in the other group intervened to resolve the conflicts.

| Quality assessment
We conducted the quality assessment in the included articles independently with the standard principles according to the Quality Assessment of Diagnostic Accuracy Study-2 (QUADAS-2) guidelines. 19 It contained eleven criteria in four parts (patient selection, index test, reference standard, and flow and timing). The criteria of QUADAS-2 are listed in Table S1. We responded with "Yes (Y), No (N), and Unclear (UC)" to assess the risk of bias and responded with "Low (L), High (H), and Unclear (UC)" in the concern of applicability. Then, the quality assessment form was filled with relevant data according to the included articles. We then discussed the result, and the third researcher resolved the dissents. Then, the statistical software Review Manager 5.3 was utilized to generate the quality plot.

| Statistical analysis
Based on the data we extracted earlier, Meta-DiSc 1.4 was employed to analyze the data and generate the results pertaining to sensitivity, specificity, negative and positive likelihood ratio (LR), diagnostic odds ratio (OR), and SROC. 20 The pooled data were discussed to determine the accuracy of LAMP for the detection of norovirus.
Heterogeneity analysis was discussed from the threshold and nonthreshold effects. The source of the heterogeneity was explored through the following operations: visual inspection of forest maps to observe the deviation and the inconsistency in the above-combined results and analysis of correlation index 21,22 ; a large deviation and inconsistency between the studies indicated a possible source of heterogeneity. The random-effects model was used to analyze the accuracy of the diagnostic method through presenting forest spots if large heterogeneity was found. Deeks' funnel plot symmetry tests were conducted by Stata 15.0 to check the potential publication bias. 23

| Database search results
After the previous detailed database search, a total of 202 publications were initially retrieved. The number then decreased to 99 after excluding the duplicates. On this basis, we excluded 68 studies after screening the abstracts and titles. After a full-text review, we excluded 27 articles. Ultimately, eight articles were identified for inclusion. [24][25][26][27][28][29][30][31] The reasons for specific exclusion are shown in Figure S1.

| Characteristics of eligible studies
Based on our strategy of data extraction, eleven sets of data extracted from the eight studies were included for meta-analysis. The detailed characteristics of the included studies are summarized in Table 1. All articles were prospective studies and published between 2008 and 2017. LAMP detected in GI or GII or both of genogroup of norovirus, the results were compared with RT-PCR. Among them, all articles reported data from stool samples; one collected vomitus samples, and one collected the samples from the anal swab, peripheral water, and barreled water. A total of 1553 samples were used to evaluate the diagnostic performance of LAMP.

| Quality assessment
To ensure scientific rigor and objectivity, two evaluations were conducted at different times. The quality assessment of the eight included articles is catalogued in Table

| Threshold effect analysis
The Spearman correlation coefficient is a nonparametric measure to evaluate the correlation between two statistical variables. If the Spearman correlation coefficient is more than 0.6, the possibility of threshold effect is indicated. Moreover, typical "shoulder-arm" patterns indicate the presence of threshold effect. 20  as we observe in Figure 8, no "shoulder-arm" distribution was seen.
Thus, we concluded that no threshold effect existed among the articles included.

| Non-threshold effect analysis
Quantitative indicators of heterogeneity were judged by the inconsistency index (I-square) automatically generated by Meta-DiSc software. The inconsistency index was interpreted in the handbook of heterogeneity. 32 The ratios are plotted with the forest map with a random pattern, and the result is shown in Figure 8 with the following values: Cochran's Q = 14.91, p = 0.1352 (p > 0.05), and inconsistency = 32.9% (inconsistency < 50%).
This indicated that the heterogeneity originating from the nonthreshold effect was low.

| Merge analysis results
With using the random-effects model, the pooled data to evaluate the diagnostic accuracy of LAMP are shown in Figures 3-7.
LAMP technology was utilized to detect norovirus, whose pooled sensitivity, specificity, positive LR, negative LR, and their 95% CI

| Publication bias
Deeks' funnel graph asymmetry was drawn, and the result, p = 0.02 (p < 0.05), is shown in Figure 9, which indicated a low potential for publication bias in the present study. However, the small number of studies included in the analysis resulted in a low capacity to detect bias.

| DISCUSS ION
Human norovirus is the leading cause of foodborne illness globally, which has caused a considerable public health and economic burden. 33 Complex pathogenesis, different susceptibility, and difficult culture methods in vitro limit the research, diagnosis, and prevention of viruses. 33  LAMP is a novel, rapid, specific, sensitive, and simple nucleic acid amplification technique located in six regions in the target gene using four primers and amplified RNA at a constant temperature. The most significant one is its rapidity; LAMP allows immediate diagnosis in just 30-50 min. 36 Recently, LAMP has been utilized for the rapid detection of some pathogens, and their results showed high accuracy and availability in the diagnosis of the corresponding pathogens. [37][38][39] In the detection of norovirus, LAMP with hydroxynaphthol blue dye could be observed visually through the change of color; therefore, the instrument of molecular electrophoresis is not required. 27 Third, the number of the included articles was comparatively small.
There were only eight articles to include and evaluate. Fourth, some articles included were relatively old. Half of the included articles were published before 2012. Finally, the existence of generic quality in the included studies could result in the rise of heterogeneity.
In conclusion, the study illustrates that LAMP has high sensitivity and specificity in the detection of norovirus. LAMP proves feasible and promising in the rapid diagnosis of norovirus infection. The results of this study provided information for finding an efficient method for the clinical detection of norovirus. However, further extension of this approach is encouraged to ensure the accuracy and practicability of this hopeful test in the future.

ACK N OWLED G EM ENTS
We show our sincere gratitude to the unity and cooperation of all members and all teachers and editors in this study for their direct and indirect help to us.

CO N FLI C T O F I NTE R E S T
The authors guarantee that there were no competing interests.