Chinese Medical Journal 2010;123(9):1201-1205
Diagnostic value of Epstein-Barr virus capsid antigen-IgA in nasopharyngeal carcinoma: a meta-analysis
LI Shan, DENG Yan, LI Xi, CHEN Qiao-pei, LIAO Xiang-cheng, QIN Xue
LI Shan (Department of Clinical Laboratory, First Affiliated Hospital of Guangxi Medical University, Nanning, Guangxi 530021, China)
DENG Yan (Graduate School of Guangxi Medical University, Nanning, Guangxi 530021, China)
LI Xi (Department of Clinical Laboratory, First Affiliated Hospital of Guangxi Medical University, Nanning, Guangxi 530021, China)
CHEN Qiao-pei (Department of Clinical Laboratory, First Affiliated Hospital of Guangxi Medical University, Nanning, Guangxi 530021, China)
LIAO Xiang-cheng (Department of Clinical Laboratory, First Affiliated Hospital of Guangxi Medical University, Nanning, Guangxi 530021, China)
QIN Xue (Department of Clinical Laboratory, First Affiliated Hospital of Guangxi Medical University, Nanning, Guangxi 530021, China)Correspondence to:QIN Xue,Department of Clinical Laboratory, First Affiliated Hospital of Guangxi Medical University, Nanning, Guangxi 530021, China (Tel: 86-771-5356052. Fax:86-771- 5356052. E-mail:firstname.lastname@example.org)
Background Non-invasive nasopharyngeal carcinoma (NPC) screening usually involves serological testing for the presence of IgA antibodies to Epstein-Barr virus (EBV) capsid antigen (VCA). The present meta-analysis determined the accuracy of VCA-IgA in the diagnosis of NPC.
Methods A systematic review of studies was conducted and data on the accuracy of VCA-IgA concentrations in the diagnosis of NPC were pooled using random effects models. Receiver operating characteristic curves were used to summarize the overall test performance.
Results Twenty studies met the inclusion criteria for the meta-analysis. The summary estimates for VCA-IgA in the diagnosis of NPC were: sensitivity 0.91 (95% confidence interval (CI): 0.90–0.92), specificity 0.92 (95% CI: 0.92–0.93), positive likelihood ratio 31.65 (95% CI: 10.99–91.15), negative likelihood ratio 0.10 (95% CI: 0.07–0.13) and diagnostic odds ratio 414.59 (95% CI: 174.96–982.42). The area under the summary receiver operating characteristic curves was 0.98.
Conclusion The sensitivity and the specificity of serum VCA-IgA are very high, suggesting that the presence of VCA-IgA in peripheral blood is a valuable predictor for NPC.
Nasopharyngeal carcinoma (NPC) has been recognized for over 100 years, the first published report of the disease is in 1901. It is an uncommon disease worldwide except in the Inuits of Alaska, the Chinese living in Southern China, and some populations in northern Africa.1 With the recent growing popularity of concurrent chemoradiotherapy, five-year overall survival and disease-free survival for advanced NPC have exceeded 70%.2,3 Although many patients with advanced NPC have a good probability of cure with concurrent chemoradiotherapy, tumor stage at initial presentation is still a major prognosticator of patient survival.4,5 Unfortunately, most NPC patients present in advanced stages of the disease at time of diagnosis in endemic areas due to the insidious clinical course of NPC and the relatively inaccessible anatomical site of the nasopharynx. A good primary NPC screening protocol may contribute to early detection and improve treatment outcome.
Non-invasive NPC screening usually involves serological testing for the presence of IgA antibodies to Epstein-Barr virus (EBV) capsid antigen (VCA) and early antigen (EA). However the discussion on the role of VCA-IgA for the specific differential diagnosis of NPC is controversial. The present meta-analysis aimed to ascertain the accuracy of VCA-IgA in the diagnosis of NPC and to explore the reasons for the ongoing controversies regarding this issue.
Identification of studies
A comprehensive search of the Chinese Biomedical Literature Database (CBMdisc) (1979–2008), the database of National Knowledge Infrastructure (CNKI) (1979–2008) and full paper database of Chinese Science and Technology of Chongqing (VIP) (1989–2008). The search terms were “nasopharyngeal carcinoma”, “NPC”, “Epstein-Barr virus”, “EBV”, “virus capsid antigen”, “VCA-IgA”, “sensitivity and specificity”, “predictive value”, “likelihood ratio”, “diagnosis”, and “accuracy”. Further searches were done by reviewing the reference lists of all available primary studies manually.
This review includes the observational studies which evaluated clinically suspected nasopharyngeal carcinoma through the diagnostic test of VCA-IgA levels. The diagnostic test was VCA-IgA levels with a cut-point of antibody titer ≥ 1:5, and the results of the diagnostic test were compared with the result of the histological analysis of standard paraffin-embedded sections. A VCA-IgA level diagnosis was considered correct if it is the same as the result of the paraffin section.
Studies were included in the meta-analysis if they provided both the sensitivity (true-positive rate) and specificity (true-negative rate) of VCA-IgA for the diagnosis of NPC, or when necessary data were provided in a scatter plot form, allowing test results to be extracted for the individual study subjects. We excluded studies that had less than 10 specimens, as well as studies that lacked the necessary data to construct 2 × 2 contingency tables. Conference abstracts and letters to journal editors were also excluded because they contained limited data. The studies were reviewed independently by two investigators (LI Shan and DENG Yan). Disagreements on study inclusion or exclusion were initially solved by consensus. Studies based on the same study participants were discussed, and only the best-quality studies were used.
Data abstraction and quality assessment
All of articles were assessed independently by two reviewers. The reviewers were blinded to publication details, any disagreement was resolved by consensus. For each study, the following individual data were extracted: the author, publication year, test method, participant characteristic, sensitivity and specificity. The methodological quality of the studies was assessed using guidelines published by the STARD (Standards for Reporting Diagnostic Accuracy, maximum score 25)6 and the QUADAS (Quality Assessment for Studies of Diagnostic Accuracy, maximum score 14).7
We used standard methods recommended for meta-analyses of diagnostic test evaluations.8 Analyses were performed using Stata 10.0 and Meta-DiSc 1.4. For each study, sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), and diagnostic odds ratio (DOR) were calculated.
The summary receiver operator characteristic (SROC) curves were constructed to illustrate the performance of VCA-IgA in the diagnosis of NPC using the area under the curve (AUC) value. In this meta-analysis, all of the outcomes were reported using random-effects models to allow for interstudy variability. Cochrane Q test and I2 test for heterogeneity were used to assess between-study heterogeneity. Statistically significant heterogeneity was considered at P <0.05 and I2 >50%.9
To assess whether study quality affected the diagnostic ability of VCA-IgA, STARD and QUADAS scores were included as covariates in a univariate meta-regression analysis. According to standard methods, the relative DOR (RDOR) was calculated to analyze the change in diagnostic precision in the study per unit increase in the covariate.10 Publication bias was examined visually by inspecting funnel plots and statistically using the Egger regression model.11
Quality of reporting and study characteristics
After independent review, 20 publications on the role of VCA-IgA concentrations in the diagnosis of NPC met the criteria for inclusion were included in the analysis.12-31 They are all the Chinese articles. The twenty articles based on 4671 patients with NPC, 7663 patients without NPC, the diagnostic characteristics of these studies, along with STARD and QUADAS scores were outlined in Table.
Table. Summary data from the 20 studies of VCA-IgA in the diagnosis of NPC
Figures 1 and 2 show the forest plot of sensitivity and specificity for 20 studies using VCA-IgA in the diagnosis of NPC. The pooled summary indexes showed that the diagnostic performance of VCA-IgA was an helpful laboratory test for NPC, with sensitivity of 0.91 (95% confidence interval (CI), 0.90 to 0.92), specificity of 0.92 (95% CI, 0.92 to 0.93), PLR of 31.65 (95% CI, 10.99 to 91.15), NLR of 0.10 (95% CI, 0.07 to 0.13), DOR of 414.59 (95% CI, 174.96 to 982.42). Cochran Q values of sensitivity, specificity, PLR, NLR and DOR were 154.33 (I2 =87.7, P <0.001), 1105.38 (I2 = 98.3, P <0.001), 2117.22 (I2 =99.1, P <0.001), 183.58 (I2 =89.6, P <0.001), and 129.81 (I2 =85.4, P <0.001), respectively, indicating a statistically significant heterogeneity between studies.
Figure 1. Forest plot of estimates of sensitivity for VCA-IgA in the diagnosis of NPC. The point estimates of sensitivity from each study are shown as solid circles. Error bars are 95% CI.
Figure 2. Forest plot of estimates of specificity for VCA-IgA in the diagnosis of NPC. The point estimates of specificity from each study are shown as solid circles. Error bars are 95% CI.
The SROC curve and its AUC present an overall summary of test performance and display the tradeoff between sensitivity and specificity, each data point in the SROC plot represents a separate study. A graph of the SROC curve for the VCA-IgA showing true-positive rates against true-negative rates from individual studies is shown in Figure 3. The SROC curve yielded a maximum joint sensitivity and specificity, AUC was 0.98, and Q* pointed 0.95, indicating a high level of overall accuracy.
Figure 3. Summary receiver operating characteristic curves for VCA-IgA in the diagnosis of NPC. Each solid circle represents each study in the meta-analysis. The size of each study is indicated by the size of the solid circle.
Multiple regression analysis and publication bias
Of the 20 included studies in this meta-analysis, 11 higher quality studies (STARD ≥13) did not have a better performance than those studies with STARD<13, the coefficient and RDOR were 0.186 and 1.21 (95% CI: 0.57–2.55), respectively, with P=0.582. Similarly, the 12 studies with QUADAS ≥10 produced RDOR that were not significantly higher than those studies with QUADAS <10, the coefficient and RDOR were 1.972 and 7.19 (95% CI: 1.00–51.75), with P=0.052. These results indicated that the study quality did not substantially affect the estimate of the diagnostic accuracy of VCA-IgA.
Egger test (Figure 4) for evaluation of publication bias showed that there was significant heterogeneity for the studies of VCA-IgA in the diagnosis of NPC (P=0.022). The funnel plot (Figure 5) used to show publication bias also showed some asymmetry. These results indicate the potential for publication bias.
Figure 4. Egger’s test.
Figure 5. Funnel graph for the assessment of potential publication bias in VCA-IgA in the diagnosis of NPC.
Since EBV-specific IgAs were found to be elevated in the sera of NPC patients more than three decades ago, many IgA or IgG antibodies to EBV antigens have been suggested as potential biomarkers for NPC.32,33 Of the various EBV-associated antibodies, EBV VCA-IgA has been proven to be a useful biomarker of NPC with relatively good sensitivity and specificity.32,34
The SROC curve is a technique for fitting a mathematical model to the scatter gram plot of sensitivity against (1-specificity).35 AUC can summarize the inherent capacity of a test to discriminate the participant with disease form those without it. If a test with an AUC close to 1, which means that it is a perfect test. However the poor test usually has AUC close to 0.5.36,37 In this meta-analysis, we can find that the AUC was 0.98, it indicated that the VCA-IgA assays was a perfect test.
The DOR is relate to different combinations of sensitivity and specificity, which describes the odds of positive test results in participants with the disease compared with the odds of positive test results in participants without the disease.38 The value of a DOR ranges from 0 to infinity, with higher values indicating higher accuracy the test performed. In the present meta-analysis, we found that the mean DOR values was 414.59 (95% CI, 174.96 to 982.42), indicating that VCA-IgA could be helpful in the diagnosis of NPC.
As the SROC curve and the DOR can not explain the clinical situation,37 while the likelihood ratios are considered to be more clinically meaningful. A likelihood ratio describes how many times a participant with disease is more likely to receive a particular test result than that without disease.39 The PLR greater than 10 or NLR less than 0.1 can provide convincing diagnostic evidence. In the present meta-analysis, a PLR of 31.65 suggests that patients with NPC have a 32-fold higher chance of being VCA-IgA test positive compared with patients without NPC. While the NLR was 0.10, all of these suggest a potential role for VCA-IgA determination in confirming NPC.
To assess the sources of heterogeneity in the studies, the analysis was used. As the RDOR can indicate a change in the diagnostic performance of the test in the study due to the increase of one unit of the corresponding covariable, in this meta-analysis, STARD and QUADAS scores were used in the meta-regression analysis to assess the effect of study quality on RDOR. Studies with higher quality (STARD ≥13 or QUADAS ≥10) produced RDOR values that were not significantly higher than those studies with lower quality, although the sensitivity, specificity, PLR, NLR and DOR between these studies indicated a significant heterogeneity.
The limitations of this meta-analysis are as follows. (1) All of the included studies were Chinese articles, this would lead to language bias. (2) We excluded the conference abstracts, letters to journal editors may have led to publication bias. (3) As some of the included studies published in 1980s, the quality of these studies was not good, which may influence the quality of this meta-analysis. (4) The original studies lack of the necessary data, such as the baseline character of the patient, study design, measurement techniques.
In conclusion, the presence of VCA-IgA in peripheral blood is a valuable predictor for NPC.
1. Chan KH, Gu YL, Ng F, Ng PS, Seto WH, Sham JS, et al. EBV specific antibody-based and DNA-based assays in serologic diagnosis of nasopharyngeal carcinoma. Int J Cancer 2003; 105: 706-709.
2. Lin JC, Jan JS, Hsu CY, Liang WM, Jiang RS, Wang WY. Phase III study of concurrent chemoradiotherapy versus radiotherapy alone for advanced nasopharyngeal carcinoma: positive effect on overall and progression-free survival. J Clin Oncol 2003; 21: 631-637.
3. Cheng SH, Jian JJ, Tsai SY, Yen KL, Chu NM, Chan KY, et al. Long-term survival of nasopharyngeal carcinoma following concomitant radiotherapy and chemotherapy. Int J Radiat Oncol Biol Phys 2000; 48: 1323-1330.
4. Lee AW, Sze WM, Au JS, Leung SF, Leung TW, Chua DT, et al. Treatment results for nasopharyngeal carcinoma in the modern era: the Hong Kong experience. Int J Radiat Oncol Biol Phys 2005; 61: 1107-1116.
5. Chang KP, Hao SP, Lin SY, Ueng SH, Pai PC, Tseng CK, et al. The 30-bp deletion of Epstein-Barr virus latent membrane protein-1 gene has no effect in nasopharyngeal carcinoma. Laryngoscope 2006; 116: 541-546.
6. Bossuyt PM, Reitsma JB, Bruns DE, Gatsonis CA, Glasziou PP, Irwig LM, et al. Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative. BMJ 2003; 326: 41-44.
7. Whiting P, Rutjes AW, Reitsma JB, Bossuyt PM, Kleijnen J. The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews. BMC Med Res Methodol 2003; 3: 25.
8. Deville WL, Buntinx F, Bouter LM, Montori VM, de Vet HC, van der Windt DA, et al. Conducting systematic reviews of diagnostic studies: didactic guidelines. BMC Med Res Methodol 2002; 2: 9.
9. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002; 21: 1539-1558.
10. Suzuki S, Moro-oka T, Choudhry NK. The conditional relative odds ratio provided less biased results for comparing diagnostic test accuracy in meta-analyses. J Clin Epidemiol 2004; 57: 461-469.
11. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997; 315: 629-634.
12. Tang ZH, Wu YF, Yang J, Rong BP, Xu QG, Zhao CN, et al. Determination of serum VCA IgA antibody by immunoenzymic method and diagnosis of nasopharyngeal carcinoma. Shanghai Med J (Chin) 1981; 4: 28-31.
13. Xie JW, Li ZQ. The research of EB virus VCA_IgA antibody detection in determination of the differential diagnosis of nasopharyngeal carcinoma. Chin J Cancer (Chin) 1983; 3: 40.
14. Li ZQ, Li WJ, Xie JW, Chen JJ, Zhang YX. The value of EB virus immunology in the diagnosis of nasopharyngeal carcinoma. Chin J Clin Oncol (Chin) 1983; 10: 206-210.
15. Gao DH, Xiao YK, Cao JX, Ji SS, Gao XK. Application of immunofluorescence assay and enzymatic to detect serum EB virus VCA_IgA antibody in the diagnosis of nasopharyngeal carcinoma. Sichuan Med J (Chin) 1984; 5: 325-326.
16. Kuang GQ, Huang CH. The value of EB virus VCA_IgA antibody detection in the diagnosis of nasopharyngeal carcinoma — analysis report of 1354 samples of serum detection. J Guangxi Med Univ (Chin) 1985; 2: 67-69.
17. He ZD, Du J, Chen CB, Liu XY, Lin L. The diagnostic value of serum EB virus-specific DNA enzyme in the diagnosis of nasopharyngeal carcinoma. J Clin Res (Chin) 1986; 3: 202-204.
18. Liu C, Wang TY. The value of serum EB virus detection in the diagnosis of nasopharyngealccarcinoma. Foreign Med Sci Oncol Sect (Chin) 1987; 13: 92-94.
19. Huang PS, SunYY, Chen YD, Wang Y, Shen YY. Clinical significance of EB vikus EA-IgA antibody detection in the nasopharyngeal carcinoma patients. J Fujian Med Univ (Chin) 1988; 22: 16-19.
20. Tian PP, Zhang XH, Zhang LS, Li FQ, Yin WD. The study of the relationship between EB virusVCA-IgA antibody detection and the diagnosis of nasopharyngeal carcinoma. Hebei Med J (Chin) 1990; 12: 370-371.
21. Zhou ZM, Cai WM, CaoY, Liu YY, Xu GZ, Hu YH. The evaluation of EB virus antibody detection in diagnosis and differential diagnosis of nasopharyngeal carcinoma-study on 200 cases using double-blind method. J Med Res (Chin) 1995; 24: 16-20.
22. Xiao YK, Zhang J, Lu YR, Lin P. Value of detection of serum Epstein-Barr virus IgA/VCA by immunoenzyme technique in the diagnosis of nasopharyngeal carcionma. Pract J Cancer (Chin) 1995; 10: 1-3.
23. Huang PC, Chen J, Pan QH. Value of detection of Epstein Barr virus IgA/VCA and IgA/EA in diagnosis of nasopharyngeal carcinoma. J Guangdong Med College (Chin) 1998; 16: 184-185.
24. Zhou YQ. The diagnosis significance of EBV-VCA-IgA in nasopharyngeal carcinoma. J Clin Transfus Lab Med (Chin) 2000; 2: 58.
25. Wang ML, Xu XH, Lu JS. Diagnostic value of EB virus antibody detection in nasopharyngeal carcinoma. Zhejiang Clin Med J (Chin) 2000; 2: 797-798.
26. Fang QQ. Application of EB virus detection and the needle aspiration of neck lymph node in thediagnosis of nasopharyngeal carcinoma. Fujian Med J (Chin) 2003; 25: 54-55.
27. Huang WG, Lin J, Chen RC. Detection of diagnosis about nasopharyngeal carcinoma combined with EBVCA-IgA and EBVA-IgA in serum. J Pract Med Tech (Chin) 2003; 10: 1235-1236.
28. Huang LS, Huang WC, Chen YH. Detection of diagnosis about nasopharyngeal carcinoma combined with CYFRA21-1, TSGF, and EB-VCA-IgA. J Guangxi Med Univ (Chin) 2005; 22: 233-234.
29. Huang LS, Zhu B, Chen YH, Zhao HL. Significance of combined detection of serum tumor makers in the diagnosis of nasopharyngeal carcinoma. Lab Med (Chin) 2006; 21: 472-474.
30. Li Z, Bi S, Gao YH, Yang X, Zhang GQ. To assess the value of EB virus antibody detection in the diagnosis of nasopharyngeal carcinoma. Guide China Med (Chin) 2007; 6: 79-80 .
31. Li Q. The relationship between EB virus VCA_IgA antibody and the diagnosis of nasopharyngeal carcinoma. J Hainan Med College (Chin) 2007; 13: 180-183.
32. Ho HC, Ng MH, Kwan HC, Chau JC. Epstein-Barr- virus-specific IgA and IgG serum antibodies in nasopharyngeal carcinoma. Br J Cancer 1976; 34: 655-660.
33. Henle G, Henle W. Epstein-Barr virus-specific IgA serum antibodies as an outstanding feature of nasopharyngeal carcinoma. Int J Cancer 1976; 17: 1-7.
34. Levine PH, Pearson GR, Armstrong M, Bengali Z, Berenberg J, Easton J, et al. The reliability of IgA antibody to Epstein-Barr virus (EBV) capsid antigen as a test for the diagnosis of nasopharyngeal carcinoma (NPC). Cancer Detect Prev 1981; 4: 307-312.
35. Littenberg B, Moses L. Estimating diagnostic accuracy from multiple conflicting reports: a new meta-analytic method. Med Decis Making 1993; 13: 313-321.
36. Altman DG. Relation between two continuous variables. In: Altman DG, ed. Practical statistics for medical research. London: Chapman & Hall; 1999: 277-299.
37. Deeks JJ. Systematic reviews of evaluation of diagnostic and screening tests. In: Egger M, Davey-Smith G, Altmann DG, eds. Systematic reviews in health care: meta-analysis in context. London: BMJ Publishing Group; 2001: 248-282.
38. Duval S, Tweedie R. Trim and fill: a simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics 2000; 56: 455-463.
39. Akobeng AK. Understanding diagnostic tests 2: likelihood ratios, pre- and post-test probabilities and their use in clinical practice. Acta Paediatr 2007; 96: 487-491.