Diagnostic accuracy of LAMP assay for HBV infection

Abstract Background Detection of hepatitis B virus (HBV) is vital for the diagnosis of hepatitis B infection. A novel test loop‐mediated isothermal amplification (LAMP) has been successfully applied to detect various pathogens. However, the accuracy of LAMP in diagnosing HBV remains unclear. Therefore, in the present study, the accuracy of LAMP for HBV detection was evaluated systematically. Methods Embase, Cochrane Library, and PubMed databases were searched for studies using LAMP to detect HBV. Then, two researchers extracted data and assessed the quality of literature using the QUADAS‐2 tool independently. I2 statistic and chi‐square test were analyzed to investigate the heterogeneity, and Deek's funnel plot assessed the publication bias. The pooled sensitivity (SEN), specificity (SPE), positive LR (PLR), negative LR (NLR), diagnostic odds ratio (DOR), and 95% confidence intervals were displayed in forest plots. We calculated the area under the curve (AUC) to assess the overall efficiency of LAMP for HBV detection. Results A total of nine studies with 1298 samples were finally included in this evaluation. The pooled sensitivity and specificity of HBV detection were 0.91 (95% CI: 0.89 ~ 0.92) and 0.97 (95% CI: 0.94 ~ 0.99), respectively. The PLR, NLR, and DOR were 16.93 (95% CI: 6.15 ~ 46.55), 0.08 (95% CI: 0.05 ~ 0.14), and 397.57 (95% CI: 145.41 ~ 1087.07). Besides, the AUC was 0.9872, and Deek's plot suggested that there existed publication bias in the studies. Conclusion Compared with PCR, LAMP is a simple, rapid, and effective assay to diagnose HBV. However, additional evidence is essential to confirm that LAMP can replace other methods in diagnosing HBV infection.


| INTRODUC TI ON
According to the report of the World Health Organization (WHO) in 2015, the hepatitis B virus (HBV) infected 275 million people worldwide and presented a threat to global health. The number of deaths caused by hepatitis B was 887 000. A total of 1.1 million new infections were detected in 2017. The region with the highest prevalence of hepatitis B was in the Western Pacific and Africa, with adult infection rates 6.2% and 6.1%, respectively. 1,2 HBV is a circular, 3.2 kbp partially double-stranded DNA with eight genotypes (A-H). 3,4 The transmission of the virus occurs mainly through blood or motherto-child. 5,6 HBV is one of the leading causes of acute and chronic hepatitis and liver cancer, which is often co-infected with the hepatitis C virus and human immunodeficiency virus. 7,8 And the gene expression of TLR1, TLR6, and NEAT1 may attribute to the innate immune response to enhance the chronic infection. 9 Some studies have determined that the HBV level was associated with HBeAg. 6 Therefore, a highly accurate method for the detection of HBV is essential for the early diagnosis, treatment, and prognosis.
Various techniques, such as enzyme-linked immunosorbent assay (ELISA) and quantitative PCR (qPCR), have been widely used to detect the hepatitis B surface antigen (HBsAg) and HBV-DNA. [10][11][12][13][14] Jean et al (2018) determined the sensitivity of BIOSYNEX IMMUNOQUICK ® RDT for HBsAg detection was 78%, higher than the previous report. 5 Matsuo et al (2017) applied "Lumipulse HBsAg-HQ" to detect the HbsAg and its variants, finding that HBsAg-HQ assay had high sensitivity and lower LOD to detect HBV. 15 However, the sensitivity of the immunization method was not high in screening hepatitis B of early stage, while the sensitivity of the PCR technique to detect HBV-DNA of acute infection was higher. 16,17 PCR-based detection has certain advantages in sensitivity, pollution control, and virus quantity test. Nevertheless, these time-consuming and expensive methods require high personnel and equipment. Thus, an alternative approach, loop-mediated isothermal amplification (LAMP), has been under intensive investigation. LAMP was developed by Notomi et al (2000) 18 and successfully applied for the detection of HBV and other pathogenic microorganisms. [19][20][21][22] It has high accuracy for the detection of microorganisms.
For example, the LAMP can detect NDM-1 gene with high specificity in 45 minutes. 23 However, LAMP, combined with other technologies, can improve the efficiency of diagnosis. To identify the viable bacteria, Wu et al (2017) 24 applied an RT-LAMP assay to detect the 16S rRNA of Mycobacterium tuberculosis and found that RT-LAMP was 10-fold higher in sensitivity than that of the LAMP. Four specific primers are designed to amplify the six particular regions of the target sequence. The Bst DNA polymerase is used for strand displacement-DNA synthesis. LAMP can be performed at about 65°C in a water bath for 60 minutes. The naked eyes or amplification curve observed the reaction.
In recent years, the detection of HBV by LAMP has been established, but the overall evaluation has not been studied thoroughly. Therefore, we combined the previous research data and made a systematic analysis to reveal the accuracy of LAMP in detecting HBV.

| Electronic searches
We conducted this systematic evaluation according to the standard guidelines. [25][26][27][28] The search using databases such as PubMed, Embase, and Cochrane Library retrieved relevant studies published from January 1, 2000, to date without language restriction. The

| Study screening and selection
Two investigators (Chu-Mao Chen and Li-Juan Wu) independently checked the title and abstract to screen out the relevant literature; any discrepancies were settled by a third investigator (Tian-Ao Xie).
The studies that met the eligibility criteria were included in this evaluation.

| Data extraction
The necessary information was extracted independently by two investigators from each of the included studies, such as the year of publication, reference method, and the number of samples.
Additionally, the diagnostic characteristics of LAMP, such as TP, FP, TN, and FN, were extracted. Then, we reviewed the information obtained by two investigators, and the disagreement was settled by a third investigator.

| Quality assessment
Two investigators assessed the quality of the included studies independently based on the QUADAS-2 standard. 29 The quality of the included studies was evaluated by two investigators. The QUADAS-2 comprised of eleven criteria in four domains (patient selection, index test, gold-standard method, and flow and timing). We used "Yes," "Unclear," and "No" to assess the risk of bias of the studies. The first three assessments considered the clinical applicability of the method.

| Statistical analysis
Statistical analysis and data synthesis were performed according to the standard method using the Meta-Disc 1.4 software. 30

| Publication bias
In a systematic evaluation, we tested the publication bias of the included studies by drawing Deek's funnel plot using Stata 12.0 software.

| Threshold effect detection
Spearman's correlation coefficient was calculated to explore the threshold effect of the included studies. A negative correlation between the sensitivity and specificity suggested a threshold effect or vice versa. 31

| Investigations of heterogeneity
We performed chi-square test and I 2 statistic to assess the heterogeneity among the included studies. 32 When a heterogeneity effect was detected, the random-effects model was selected. If the heterogeneity was small or absent, the fixed-effects model was implemented.

| Data synthesis
We used the Meta-Disc 1.4 software to merge the sensitivity, specificity, positive LR (PLR), and negative LR (NLR) using an appropriate model. The results were shown in forest plots. Also, SROC was applied to calculate the AUC. Then, DOR and AUC were used to evaluate the overall efficiency of the LAMP assay. Each item would combine the 95% confidence interval (95% CI). 33

| Literature search
Based on the current search strategy, we identified 153 related articles: 103 articles were from PubMed, 25 from Embase, and 25 from the Cochrane Library. After excluding 19 repetitive items, we screened the titles, abstracts, and full text. Finally, 125 ineligible F I G U R E 1 Flow diagram of study identification and inclusion articles were excluded according to the criteria, and nine articles were included. 19,[34][35][36][37][38][39][40][41] The process is represented in Figure 1.

| Study characteristics
The characteristic details of the articles are shown in Table 1

| Quality assessment of the study
According to QUADAS-2, 29 the methodological quality of nine articles was evaluated independently by two researchers. The criteria we used to assess the methodological quality were catalogued in Table S1. Quality assessment of the included studies was shown in Table S2. The risk of bias and applicability results are presented in Figures 2 and 3 and Table S3. In terms of patient selection, six studies were evaluated to offer a high risk of bias because they included the HBV-confirmed cases or set up a case-control trial, or did not avoid inappropriate exclusion, while two studies were evaluated as low risk of bias. Nevertheless, the remaining studies were not clear because the description of the case was inadequate. In the aspect of the index test, approximately 33% of the studies were considered as low risk of bias while the remaining were at unclear risk due to their indistinct thresholds, or the results of the gold-standard method were known when explaining the index test.

| Diagnostic accuracy
The forest plots were drawn with the random-effects model using the Meta-Disc 1.4 software. As presented in Figure 4, we obtained the following results: Cochran-Q = 4.74, P = .7847, and inconsistency = 0.0%, which indicated that the threshold effect caused no heterogeneity. Then, we merged the data to evaluate the diagnostic accuracy of LAMP (Figures 5-8). The combined sensitivity,  Figure 4, the DOR was 397.57 (95% CI: 145.41-1087.07). Thus, we adopted a random-effects model to draw the SROC (Figure 9). The results showed that the AUC = 0.9872 and Q index was 0.9519 (SE = 0.0119). Both DOR and SROC suggested that LAMP had high diagnostic accuracy for the HBV infection.

| Publication bias
Deek's funnel plot (Figure 10) was drawn using Stata 12.0. The test indicated a potential publication bias in the included studies (P = .026).

| D ISCUSS I ON
LAMP is a novel nucleic acid amplification method, which amplifies six regions on the target sequence using four specific primers and Third, the existence of general quality research led to a certain degree of heterogeneity. Finally, the detection ability of the reference method was not stronger than that of LAMP technology, which led to false-positive results, thereby underestimating the specificity of the LAMP method.
In conclusion, LAMP is a rapid, sensitive, and specific detection method, providing reliable experimental results for the diagnosis of HBV infection. Also, it improves the treatment of patients and reduces the financial burden. With the development and maturity of technology, LAMP might become a primary auxiliary diagnostic tool for HBV infection.

CO N FLI C T O F I NTE R E S T
The authors declare that there are no competing interests associated with the manuscript.