A meta-analysis on the EBV DNA and VCA-IgA in diagnosis of Nasopharyngeal Carcinoma

Objective: We conducted a meta-analysis to compare the EBV DNA and VCA-IgA in diagnosis of Nasopharyngeal Carcinoma, and provide important evidence for screening method of NPC. Methodology: Three databases, Medline (from Jan. 1966 to Jan. 2012), EMBASE (from January 1988 to Jan. 2012) and Chinese Biomedical Database (from January 1980 to Jan. 2012) were used to detect the role of EBV DNA and VCA-IgA in diagnosis of NPC. Meta-DiSc statistical software was used for analysis. Results: Twenty seven case-control and cohort studies were included in final analysis. A total of 1554 cases and 2932 controls were included in our meta-analysis. The Sensitivity specificity, positive likelihood (+LR) and likelihood negative (-LR) of EBV-DNA in diagnosis of NPC were 0.75(0.72-0.76), 0.87(0.85-0.88), 6.98(4.50-10.83) and 0.18(0.11-0.29), respectively, and they were 0.83(0.81-0.85), 0.85(0.83-0.86), 10.89(5.41-21.93) and 0.20(0.14-0.29) for VCA-IgA. The SROC for EBV DNA detection was 0.939, while this was 0.936 for VCA-IgA detection. The subgroup analysis showed EBV-DNA had larger areas under the summary receiver operator curve when compared with VCA-IgA in high quality and low quality studies. Conclusion: Our meta-analysis indicated the EBV DNA had higher sensitivity and specificity in diagnosis of NPC.


INTRODUCTION
Nasopharyngeal carcinoma (NPC) is a rare disease on a world scale, and it accounted for 0.7% of all cancers, and ranked the 23rd most common new cancer in the world. 1 However, it is endemic in some specific areas, such as in Hong Kong, and south of China. 1 The intermediate rates are observed in several indigenous populations in South East Asia and in natives of the Arctic region, North Africa and the Middle East. 1 Epstein-Barr Virus (EBV) is a well known risk factor for NPC. Patients with NPC are noted to have high levels of EBV antibodies. 2,3 The infection of EBV is not associated directly in inducing by the tumor, but infection in the healthy individuals means increased risk of cancer. [4][5][6] Several diagnostic methods are used for NPC detection, but the EBV serology examinations test IgA antibodies against viral capsid antigen (VCA) to IgA to early antigen are the most common detection methods for diagnosis of NPC. 2 This method is cheap and non-invasive, and therefore, it is acceptable for patients and could be widely used in clinics. Quantitative EBV DNA and VCA-IgA analysis has been reported to be a sensitive detection tool in diagnosis of NPC. 7,8 Recent studies indicated the cell-free EBV DNA had high detection rate in the plasma and serum among patients with NPC. 9,10 Recently several studies have showed plasma EBV-DNA and VCA-IgA level might be a sensitive and reliable biomarker for the diagnosis of NPC at a molecular level in clinical practice. 11-14 However, the there is no consensus yet which is a better test for the early diagnosis of nasopharyngeal carcinoma. Reasons may include the different sources of EBV antigens, different antibody assays and the selection of cases from different geographic origins. Therefore, we conducted a meta-analysis to evaluate which EBV serology examination had the better sensitivity and specificity in the diagnosis of NPC.

METHODOLOGY
Searching strategy: Three databases, Medline (from Jan. 1966 to Jan. 2012), EMBASE (from January 1988 to Jan. 2012) and Chinese Biomedical Database (from January 1980 to Jan. 2012), were systematically searched by using related terms ('Epstein-Barr Virus', 'EBV', 'DNA', 'VCA-IgA', 'serological test', 'nasopharyngeal carcinoma' (NPC). There was no restriction on the language of the papers. References cited in retrieved studies were reviewed for more eligible studies. The criteria used for including studies were (1) Case-control or cohort studies on the role of EBV-DNA and VCA-IgA in diagnosis of NPC; (2) identification of NPC was confirmed histologically/pathologically; (3) Available data regarding sensitivity and specificity of EBV-DNA and VCA-IgA in diagnosis of NPC; If the authors reported more than once the data on publication papers, we only included the complete data into our review. The exclusion criteria were case only study, reviews, and overlapping studies. Data extraction: Two reviewers independently reviewed the final abstracts of all potential articles, and decided one should be included into final metaanalysis. In case there was any disagreement, it was resolved by discussion. If the data were missing in the included studies, we attempted to contact the authors by emails or telephones in order to include complete data. From these finally selected studies, we included author's names, location, study type,

Characteristics of studies:
A total of 758 records were selected by searching the databases. After excluding the overlapping studies and those which were not in line with the inclusion criteria. A total of 29 studies were included and assessed for metaanalysis. After reviewing the original paper, we excluded 2 studies. Finally, 27 case-control and cohort studies were included in final analysis. A total of 2717 cases and 4085 controls were included in our meta-analysis (Table-I).
We analyzed the pooled sensitivity, specificity, positive likelihood (+LR) and likelihood negative (-LR) of EBV-DNA and VCA-IgA (Table II   EBV DNA detection was 0.939, while the SROC was 0.936 for VCA-IgA detection ( Fig. 1 and 2).
In the pooled analysis for EBV-DNA, there was significant heterogeneity across studies (p<0.05, I 2 >50%). While, no significant heterogeneity was found between studies in terms of VCA-IgA. Subgroup analysis was taken according to the quality of studies to investigate the heterogeneity within the included studies (Table-IV), which indicated studies with low quality had lower sensitivity, specificity, +LR and -LR for both EBV-DNA and VCA-IgA detection. We could find the EBV-DNA had larger areas under the summary receiver operator curve when compared with VCA-IgA in high quality and low quality studies. The subgroup analysis significantly decreases the heterogeneity among studies, with the p value of 0.12 for EBV-DNA and 0.31 for VCA-IgA methods.
A single study in our meta-analysis was removed each time to analyze the robust of the pooled results, and the results did not greatly changed (Data not shown). The Egger's test were used to assess the publication bias, and no significant publication bias was found in our meta-analysis.

DISCUSSION
Meta-analysis has been regarded as an important tool to more precisely define the effect of treatment for diseases and to identify potentially important sources of between-study heterogeneity. There is no systematic review to compare the EBV DNA and VCA-IgA in diagnosis of NPC. Only one previous     Heterogeneity is a potential problem in explaining the results of meta-analysis, and identifying the sources of heterogeneity is an important goals of meta-analysis. 39 In our study, we assessed the between-study heterogeneity by using the I2 statistic to quantify the between-study heterogeneity 39 , and the results suggested great heterogeneity between studies in terms of EBV-DNA. Therefore, we further performed subgroup analysis by risk of bias. The results showed that risk of bias was an main source of heterogeneity.
There are two possible limitations in our metaanalysis which mainly influence the explanation of the results. Firstly, there might be publication bias in our study. All the studies included into meta-analysis were published paper; however, there might be many unfavorable results which may not have been published. We plan to include more studies in clinical trials registration and paper presented in conferences. Secondly, there might be selection bias in our study. Secondly as most of the studies included the NPC cases and controls in the same hospital or places, which could influence the results of study.
In conclusion, our results demonstrated the EBV DNA and VCA-IgA detection methods had better effect in diagnosis of NPC. However, EBV DNA detection method had high accuracy in diagnosis of NPC.