Diagnostic value of fibulin-3 for malignant pleural mesothelioma: A systematic review and meta-analysis

Background Several studies have investigated the diagnostic value of fibulin-3 for malignant pleural mesothelioma (MPM), but the results were various. Therefore, we performed a systematic review and meta-analysis to evaluate the diagnostic value of fibulin-3 for MPM. Results Eight studies were included in this work. The overall sensitivity of blood fibulin-3 were 0.87 (95% CI, 0.58 – 0.97) and 0.89 (95% CI, 0.77 – 0.95), respectively. The overall sensitivity and specificity of PF fibulin-3 for MPM were 0.73 (95% CI, 0.54 – 0.86) and 0.80 (95% CI, 0.60 – 0.91), respectively. The area under curve of blood and pleural effusion (PF) Fibulin-3 were 0.94 (95% CI, 0.91 – 0.96) 0.83 (95% CI, 0.79 – 0.86), respectively. Methods PubMed and EMBASE databases were searched up to July 29, 2016 to verify studies investigating the diagnostic value of fibulin-3 for MPM. The quality of eligible studies was assessed using the revised Quality Assessment for Studies of Diagnostic Accuracy tool (QUADAS-2). The overall sensitivity and specificity were pooled using a bivariate model. Conclusion Fibuoin-3 is a useful diagnostic marker for MPM.


INTRODUCTION
Malignant pleural mesothelioma (MPM) is one of the most common cancers in asbestos-exposed individuals around the world [1,2]. Timely and accurate diagnosis of MPM can improve the outcomes of patients [3]. Currently, the diagnosis of MPM mainly relies on pleural biopsy, which is invasive and the sample error is a problem [4,5]. Therefore, developing non-invasive biomarkers for MPM diagnosis is of great value [6]. During past decades, many circulating biomarkers for MPM has been developed, such as soluble mesothelin-related peptides (SMRP) [7,8] and osteopontin [9,10]. However, the sensitivity and specificity of theses biomarkers are modest. Therefore, it is valuable to explore novel biomarkers that can improve the diagnostic value of the traditional biomarkers or replace them.
Fibulin-3 is a secreted glycoprotein that plays an important role in the regulation of cell migration and proliferation [11]. During past years, many studies have revealed that fibulin-3, either in blood or pleural effusion (PF), is a potential diagnostic biomarker for MPM [6]. However, the results from these studies were heterogeneous. Therefore, we performed a systematic review and meta-analysis to investigate the diagnostic value of fibulin-3 for MPM.

Summary of eligible studies
Eight studies were included in present systematic review and meta-analysis [12][13][14][15][16][17][18][19]. A flowchart depicting study selection is shown in Figure 1. Summary of eligible Research Paper studies is listed in Table 1. Eight studies investigated the diagnostic value of blood (three used serum [15,17,18] and five used plasma [12-14, 16, 19]) fibulin-3 for MPM and five studies investigated the diagnostic value of PF fibulin-3. The study performed by Pass et.al. [12] contained two study cohorts and thus is regarded as two independent studies. The sample sizes ranged from 36 to 228. Components of control in eligible studies were various, including asbestos-exposed persons [12,13,18,19], patients with pleural effusion [13,16,19] or metastatic pleural malignancy [15], healthy controls [17] or patients with extrapleural pneumonectomy [14]. One study did not report reference standard used for MPM diagnosis [12], and one study [13] used biopsy and followup to diagnose MPM. The remaining studies used biopsy to as reference standard. One study [14] was retrospective design and two [15,19] did not report the type of design; and the remaining studies were prospective design.

Quality assessment of eligible studies
Quality assessment of eligible studies is listed in Table 2. The patient selection domain of all eligible studies was labeled as high because lacking of uniform including and excluding criteria, and the study cohort in each eligible studies was not enrolled consecutively. Index domain in all eligible studies, except one [13], was labeled as unknown because whether index test was performed in a blind manner was not reported. Reference standard domain in one study [12] was labeled as unknown because the diagnostic criteria for MPM was not reported. Flow and timing domain of four studies was labeled as high because of the differential verification bias [17,18], partial verification bias [12] or disease progression bias [13]. Table 3 lists the diagnostic value of fibulin-3 in each eligible studies. All studies used ELISA to determine fibulin-3. Figure 2 is a forest plot for blood fibulin-3. The overall sensitivity and specificity of blood fibulin-3 for MPM were 0.87 (95% CI, 0.58 -0.97) and 0.89 (95% CI, 0.77 -0.95), respectively. Significant heterogeneity was observed for both sensitivity and specificity, with   [14] High Unknown Low Low High Low Low Agha [15] High Unknown Low Low Low Low Low Elgazzar [16] High Unknown Low Low Low Low Low Kaya [17] High Unknown Low High High Low Low Demir [18] High Unknown Low High Low Low Low Napolitano [19] High Unknown Low Low Low Low Low PF Pass [12] High High Unknown High Low Low Unknown Creaney [13] High Low Low High Low Low Low Kirschner [14] High Unknown Low Low Low Low Low Agha [15] High Unknown Low Low Low Low Low Elgazzar [16] High Unknown Low Low Low Low Low  Table 4 lists the results of subgroup analysis. Type of data collection (prospective or others), matrix used for fibulin-3 measurement (serum or plasma) and the components of controls (asbestosexposed individual only or others), were not the sources of heterogeneity. The overall sensitivity and specificity of PF fibulin-3 for MPM were 0.73 (95% CI, 0.54 -0.86) and 0.80 (95% CI, 0.60 -0.91), respectively. The I 2 for sensitivity and specificity were 91 and 92 respectively, indicating that great heterogeneity was exist among eligible studies.

DISCUSSION
To the best of our knowledge, this is the first metaanalysis investigating the diagnostic value of blood and PF fubulin-3 for MPM. The major findings of present study are as follows. First, both blood and PF fibulin-3 were useful diagnostic markers for MPM. Second, the available studies have some design weakness and further well-designed studies are needed to rigorously evaluate the diagnostic value of fibulin-3. Third, there was no publication bias among all available studies, indicating that the results of present work are reliable.
To present, many diagnostic biomarkers for MPM has been developed. Among these biomarkers, SMRP and osteopontin are the most promising ones [12,20]. Evidence from meta-analysis revealed that the diagnostic sensitivity and specificity for SMRP were 0.61 and 0.87, respectively [7]. For osteopontin, a meta-analysis also revealed that its sensitivity and specificity were 0.65 and 0.81, respectively [10]. Our study revealed that the diagnostic specificity of blood fibulin-3 was 0.89, which was comparable to those of SMRP and osteopontin. However, the sensitivity of blood fibulin-3 was 0.87, which was obviously higher than those of SMRP and osteopontin. Therefore, blood fibulin-3 represents a promising diagnostic marker for MPM, and it may have a potential to replace SMRP or osteopontin in MPM diagnosis. Further studies designed in a head-tohead comparison manner are needed to clarify whether the diagnostic value of fibulin-3 is superior to those of SMRP and osteopontin.
AUC of sROC is an index determining overall diagnostic value of a test [21,22]. Our study revealed that the AUCs for blood and PF fibuin-3 were 0.94 and 0.83, respectively, indicating that both PF and blood fibulin-3 are useful for MPM diagnosis. Besides, previous meta-analysis also show that the AUCs for SMRP and osteopontin were 0.81 and 0.83, respectively, also supporting our hypothesis that the diagnostic value of fibulin-3 is superior to those of SMRP and osteopontin.
The sources of PF fibulin-3 are largely unknown. Three studies have investigated the correlation between    AEP, asbestos-exposed person; CI, confidence interval.
circulating and PF fibulin-3 [12,13,15], and two failed to observe a positive correlation between circulating and PF fibulin-3 [12,13]. These results indicate that PF fibulin-3 is not derived from circulating fibulin-3, and measuring fibulin-3 in PF may yield additional diagnostic value. However, in a study performed by Agha et al. [15], there was a good relationship between serum and PF fibulin-3.
Reasons for inconsistent finding across these three studies are unknown. Further studies are needed to elucidate the basis of the discrepancy. An exploration of the sources for heterogeneity, rather than pooling the results of all eligible studies, is an important goal of meta-analysis. Because great heterogeneity was observed across all eligible studies investigating diagnostic value of blood fibulin-3 for MPM, we performed a subgroup analysis to explore the sources of heterogeneity. We found that some of the design characteristics, including type of data collection, matrix used for fibulin-3 measurement and the components of controls, were not the sources of heterogeneity. Future studies with more eligible studies are needed to explore the sources of heterogeneity.
To facilitate more well-designed future studies on this topic, some of the methodological weakness of the available studies should be noted. The major design weakness of available studies was two-gate design [23], which can result participant selection bias. The subjects in all studies were not enrolled according to a pre-designed inclusion and exclusion criteria. That means, the subjects enrolled in these studies may not be representative of the target population in whom MPM is suspected. Therefore, the conclusions of available studies should be interpreted with caution.
Taken together, present study indicated that fibulin-3 was a useful diagnostic marker for MPM. Due to the small number of eligible studies, and all of the eligible studies have higher risk for subject selection, further welldesigned studies are needed to rigorously evaluate the diagnostic value of fibulin-3 for MPM.

Database and literature retrieve
This systematic review and meta-analysis was performed in accordance with preferred reporting items for systematic reviews and meta-analyses (PRISMA) guideline [24]. Two investigator independently searched PubMed and EMBASE for potential eligible studies. The last search date is July 29, 2016. The searched algorithm used for searching PubMed was "(Fibulin-3 OR EFEMP1 protein, human[nm] OR "Fibulin 3") AND mesothelioma". Similar search strategy was used for searching EMBASE. A manual search was also performed by reviewing references listed at the end of retrieved publications.

Inclusion and exclusion criteria
Inclusion criteria of this systematic review and meta-analysis were: (1) studies investigating the diagnostic value of fibulin-3, either in blood or PF, for MPM; (2) A 2 by 2 table can be constructed using sensitivity, specificity and sample size reported, or using data presented in the scatter plot. Conference abstracts and animal studies were excluded. Studies with sample size less than 10 were excluded because studies with small sample sizes can yield bias.
The study selection process was performed by two independent investigators. In the first round, the titles and abstracts of retrieved publications were screened and the irrelevant studies were excluded. In the second round, a full text reviewing was performed to select eligible studies for the remaining studies. Disagreements were resolved by consensus or full text review.

Data extraction and quality assessment
Two investigators independently extracted data from eligible studies. Following data were extracted: sample size, publication year, sources of participants, components of control, reference standard used for MPM diagnosis, type of design (prospective or retrospective) and fibulin-3 measurement methods. For each eligible studies, a 2 by 2 table, which consisted of true positive (TP), false negative (FN), false positive (FP) and true negative (TN), was constructed.
The revised Quality Assessment for Studies of Diagnostic Accuracy tool (QUADAS-2) [25] was used to assess the quality of eligible studies. Any disagreement in quality assessment was resolved by consensus. The corresponding authors of the eligible studies were not contacted for unknown information regarding study design.

Statistical analysis
The overall sensitivity and specificity of fibulin-3 for MPM diagnosis were pooled using a bivariate model [26]. The summary receiver operating characteristic (sROC) curve was constructed to depict the overall diagnostic value of fibulin-3 [27]. The funnel plots and the Deeks's test were used to test publication bias [28]. All analyses were performed in STATA 13.0 (Stata Corp LP, College Station, TX) and the midas command was used for all statistical analyses.