Diagnostic value of D2-40 immunostaining for malignant mesothelioma: a meta-analysis

Malignant mesothelioma (MM) has become a global disease burden for its rising incidence and invariable fatality. D2-40 has been widely used as an immunostaining marker of diagnosing MM, while its diagnostic value has not yet been evaluated. Our study aimed to assess the overall accuracy of D2-40 immunostaining for diagnosing MM through a meta-analysis. A total of 22 studies with 2,264 participants were identified from PubMed, EMBASE, Web of Science, Scopus and the Cochrane database. The pooled sensitivity and specificity of D2-40 for MM was 0.86 (95% CI: 0.84–0.89) and 0.77 (95% CI: 0.74–0.79), respectively. The area under the summary receiver operating characteristic curve is 0.93, with a diagnostic odds ratio 40.37 (95% CI: 19.97–81.61). None of the study variates was found to be a source of heterogeneity after meta-regression analysis. In conclusion, D2-40 immunostaining may not give sufficient evidence by itself to diagnose MM and should be in combination with other markers to improve the accuracy of diagnosis.


INTRODUCTION
Malignant mesothelioma (MM) is a highly aggressive carcinoma usually involving the pleura or peritoneum. Between 1994 and 2008, a total of 92,253 mesothelioma deaths has been reported to the World Health Organization from 83 countries, and the crude and ageadjusted mortality rates of mesothelioma were estimated as 6.2 and 4.9 per million population, respectively [1]. Moreover, an incidence peak was expected around 2020 according to the widespread exposure to asbestos [2]. Unfortunately, the patients suffering from MM showed poor response to the drug, and their median survival period was only 8 months (range 1~69 months) [3].
The diagnosis of carcinoma is usually based on histopathological examination on tissue or cytological specimens. Because morphological features of MM mimic those of a variety of other carcinomas, immunohistochemical markers have been used to improve the diagnostic performance, as recommended by International Mesothelioma Interest Group [4]. Whereas, several followup studies reported that the misdiagnosis rate of MM was high (10% ~44%) [5][6][7], even much higher (65%) in its pleural localization [8]. So, it is necessary to evaluate the diagnostic accuracy of the markers.
D2-40 is a commercial monoclonal antibody directed against the M2a antigen, which is associated with testicular and extratesticular seminomas and intratubular germ cell neoplasms [9]. Also, D2-40 is a marker for lymphatic endothelium and derived carcinomas [10]. In the past decade, D2-40 has been widely used as an immunostaining marker of diagnosing MM [4,[11][12][13][14]. However, the studies about its diagnostic accuracy obtained inconsistent results. Therefore, our study aimed to assess the overall accuracy of D2-40 immunostaining for diagnosing MM through a meta-analysis.

Literature search and study selection
We conducted an independent literature search to identify the relevant studies in PubMed, EMBASE, Web of Science, Scopus and the Cochrane database up to December 31, 2016. The search terms were "Mesothelioma", and "D2-40", and "sensitivity or specificity or accuracy". The reference lists of eligible studies were also manually searched to find the potential studies.
The included studies should meet all the following criteria: (1) it was an original study using D2-40 immunostaining to diagnose MM; (2) there was a diagnostic standard for MM; (3) there were sufficient data to generate a 2 × 2 table for calculating sensitivity and specificity; (4) it was published in English; (5) it involved at least 10 participants for case or control group.
Two independent researchers (CH and BW) screened and selected the eligible articles. Any disagreement between the two researchers was resolved by consultation with a third researcher (YCS).

Data extraction and quality assessment
Two researchers (CH and BW) manually extracted the following data from each study: author; publication year; research country; participant; specimen; specimen preparation method; D2-40 antibody (clone, dilution, source); cutoff of positive immunostaining; the number of true-positive, false-positive, true-negative, and false-negative results; study design; and blinding. At the same time, the methodological quality of these studies was evaluated using the Quality Assessment for Diagnostic Accuracy Studies (QUADAS) score [15]. The discrepancies were resolved through consultation with a third researcher (YCS).

Statistical analyses
The following parameters of diagnostic accuracy, together with their 95% confidence interval (95% CI), were calculated: sensitivity, specificity, positive/negative likelihood ratio (PLR/NLR), and diagnostic odds ratio (DOR). Forest plots for sensitivity and specificity were constructed. Summary receiver operating characteristic (SROC) curve was generated, and area under the cure (AUC) was calculated to assess the overall diagnostic performance. Publication bias was tested using Deeks' funnel plots [16].
Potential heterogeneity among included studies was evaluated using the I 2 inconsistency test. I 2 > 50% suggested substantial heterogeneity, which was subsequently analyzed through a meta-regression analysis to determine possible sources of heterogeneity among the studies.
All the statistical analysis was completed by Meta-DiSc XI (Cochrane Colloquium, Barcelona, Spain) and STATA 12.0 (Stata Corporation, TX, USA) software. All statistical tests were two-sided, and statistical significance was set at P < 0.05.
After subgroup analysis, diagnostic indices of D2-40 for each subtype of MM (epithelioid, or biphasic, and sarcomatoid) were shown in Figure 3B-3D and Table 3. Parameters of D2-40 for pleural MM were also listed in Table 3.

Publication bias
The slope coefficient of Deeks' funnel plot was associated with a P value of 0.48, suggesting symmetry in the data and low likelihood of such kind of bias among the included studies (Figure 4).

DISCUSSION
The search for biomarkers, a cost-effective means of MM management, has been on-going for the past years [37]. D2-40 has emerged as a promising candidate, but the results of the studies remain controversial [20,33,36]. In our study, we determined the overall accuracy of D2-40 immunostaining in diagnosing MM through a metaanalysis.
Our meta-analysis found the pooled sensitivity and specificity was 0.86 and 0.77 respectively, suggesting a rate of missed diagnoses (14%) and misdiagnosis (23%) of D2-40 for diagnosing MM. The pooled DOR of D2-40 for MM was 40.37, which indicates a modest level of overall accuracy. The pooled PLR value of 5.15 suggests that the patients with MM have an approximately 5-fold higher chance of giving a positive D2-40 result than do patients without MM. At the same time, the pooled NLR was 0.18, indicating that a negative D2-40 result is 18% likely to be a false negative, which is not low enough to rule out MM. Thus, these findings suggested that D2-40 should be combined with other markers to improve the diagnostic accuracy. For example, several studies reported that D2-40 showed low specificity in distinguishing MM from ovarian carcinomas [17,20,22]. If the differential diagnosis of peritoneal MM from ovarian carcinomas is required, clinical history of the patient should be an important factor for evaluation. At the same time, the ovarian carcinoma markers, estrogen receptor etc., should be added into the immunohistochemical panel for avoiding a false-positive diagnosis of MM [38]. MM has distinctive histological subtypes: epithelioid (60%~80%), sarcomatoid (< 10%), and biphasic (10%~15%, composed of both epithelioid and sarcomatoid components) [39]. Our study further determined the diagnostic performance of D2-40 for these subtypes and found the performance of D2-40 for diagnosing epithelioid MM and biphasic MM is better than that for sarcomatoid MM. It indicated the reactivity to D2-40 antibody differed among these subtypes of MM. A relatively low pooled sensitivity (0.65, 95% CI: 0.56-0.73) for sarcomatoid MM was obtained in our meta-analysis. However, Chirieac LR [40] have reported that all the 24 (100%) cases of sarcomatoid MM were positive for D2-40 staining in their study, which was not included in our meta-analysis for lack of control group. Further work should be performed to elucidate this issue.
In diagnostic practice, pleural MM can easily be confused with metastatic carcinomas with the pleura involvement. Our meta-analysis has performed a subgroup analysis of pleural MM. We found that the sensitivity, specificity, and AUC of D2-40 for diagnosing pleural MM were 0.85, 0.81 and 0.93, respectively. However, interpretation should be with caution when differentiating pleural MM from lung small cell carcinomas (SCC), because it was reported that D2-40 can stained 77% of lung SCC [5]. Recently, novel markers, such as BAP1 [5], MUC4 [35], fibulin-3 [41], have been used to improve the diagnostic accuracy of D2-40 for pleural MM.
Compared with histological diagnosis, cytological diagnosis of MM is minimally invasive and more easily performed. Because only 4 studies using cytology of effusions to diagnose MM can be included in our metaanalysis, the accuracy of diagnosing MM using cytology is hard to be evaluated accurately now. But the difference of   biological behavior among the subtypes of MM should be considered when using cytological diagnosis. Epithelioid MM cells readily shed into the pleural or peritoneal space, which can be identified from pleural or peritoneal effusions [42]. While, sarcomatoid MM cells generally do not shed into the pleural or peritoneal space, which leads to poor performance of cytological diagnosis [42].
Thus, cytology-only approach should be not suitable for diagnosing sarcomatoid MM and biphasic MM. Because both benign and malignant mesothelial cells can react to D2-40 antibody [22,43], it is needed to distinguish MM from reactive mesothelial hyperplasia in diagnostic practice. Several aspects (such as cellularity, papillae, zonation, growth pattern, vascularity and stromal    invasion) should be considered [4,39,44]. A new tool, combining molecular data and computational analysis, has been applied for this kind of differential diagnosis [45].
In addition, the operation procedure of D2-40 immunostaining was not uniformed in the included studies. First, though all the included studies used D2-40 clone, its dilution factors ranged from 1:25 to 1:200. Second, cut-off value of positive staining is not consistent among the included studies. According to our experience in D2-40 immunostaining (D2-40 clone, 1:100, Dako, membrane staining), optimal condition, standard operation procedure, and reliable quality control are very important for obtaining satisfactory immunostaining results.
The findings of this meta-analysis should be interpreted with caution because of several limitations. We applied strict inclusion criteria, which may be good for reducing selection bias, but led to only a small set of studies could be included. In addition, we observed the substantial heterogeneity across the included studies, but we were unable to identify its source using metaregression analysis. Future work should aim to decrease the risk of bias.

CONCLUSIONS
In summary, our findings indicated that the accuracy of D2-40 immunostaining may be not enough to diagnose MM alone. The results of D2-40 immunostaining should be interpreted in combination with those of other markers. Novel promising markers are also needed to be explored to improve the diagnostic accuracy.

Authors ̕ contributions
Chao He and Bo Wang conceived, designed the study and wrote the paper. Chun Wan and Ting Yang performed data analysis and revised the manuscript. Yongchun Shen is guarantor of the paper, taking responsibility for the integrity of the work as a whole, from inception to published article.