Description of studies
A detailed PRISMA ﬂowchart of the study identification, screening, and exclusion process was shown in Fig. 1. The primary manual search yielded 426 potentially eligible literatures through searching of electronic databases and 1 record by manual search. After excluding duplicate studies (198 studies) , 229 publications were left for screening, of which 102 records were excluded according to the inclusion and exclusion criteria from database searching. Then, the remaining 127 articles were further evaluated by abstract reviewing, and 67 studies were discarded either due to cell or animal studies data. After carefully reviewing titles and abstracts, 60 studies were considered in full-text articles assessed for suitability. 20 studies were precluded for obvious irrelevance, 16 studies were precluded for other cancer studies, and 12 studies dismissed due to no related essay (Also see Additional file 1: Table S2). Finally, 12 studies were presented in this meta-analysis [43-53].
Characteristics of studies
The demographic information of all relevant studies was detailed in Table 1. According to this table, a total of 12 studies with 978 MM patients were included in this systematic review and meta-analysis, between 1999 and 2017. Most of the studies were conducted in people of the Asian race, tracked by 7 studies (58.4%) [47, 48, 50-54], 4 studies (33.4%) in European countries [44, 46, 49, 55], one study in USA (8.2%) , and without any studies from African populations. Gender subgroups among 978 patients, 377 and 307 patients were male and female, respectively. The main clinicopathological characteristics of the included studies are shown in Table 2. More than 80% of the MM patients were diagnosed by histopathological examination. PAS combined-staining with endothelial markers (CD31 or CD34) is a commonly used method for identification of tumor VM in paraffin-embedded tissue specimens (66.7%) in 8 studies [44, 48, 50-55] as well as PAS staining in 4 studies [43, 46, 47, 49]. Also, significant predictors of VM+ in both adjusted and unadjusted analyses were Clark level IV/V (84.4%). Finally, eleven studies reported the relationship between VM and clinicopathological parameters regarding OS [43-54], with the follow-up period ranged from 39 to 480 months.
Table 1 Demographics information of included studies
Table 2 Main clinicopathological and vasculogenic mimicry characteristics of all relevant studies
All 12 papers were methodologically essayed by NOS and QUADAS-2 quality evaluation standards of the Cochrane Reviewer handbook. Both systems’ tools focus on the study as dependent on methodology. Overall, the average NOS score was approximately 7.4 out of 12, which were classified in the high quality relatively. For each study, NOS score is sorted in the Table 1. Furthermore, QUADAS-2 results confirmed that significant bias was not presented in current meta-analyses. Details of the quality evaluation of eligible studies according to the NOS score were summarized in the Additional file 1: Table S3. The reviewers' decisions about each risk of bias and applicability concerns graph presented as percentages across selected studies. Figure S1 shows all parameters of QUADAS-2 assessment individually (Additional file 2: Figure S1). In this study, no significant bias and applicability concerns were found in all selected studies.
Outcome of the meta-analysis
The association between VM+ and overall survival of MM patients was identified using the pooled proportions test method. We used a random effect approach because the heterogeneity of the overall prognosis was relatively high; which is shown across the study (I2 = 79.8, P-value < 0.001). Based on heterogeneous cross of 12 studies, VM was associated with poor prognosis in 38% of MM group compared to the VM-group (P = 0.35, 95% confidence intervals (95% CIs): 0.27-0.42, P-value < 0.001). Therefore, these results suggested that VM+ indicated a poorer prognosis for MM patients (Fig. 2).
To assess the heterogeneity from threshold effect, we conducted analysis of diagnostic threshold with the spearman correlation coefficient. The forest plots of pooled sensitivity, specificity, with their 95% CIs for individual studies are shown in the Fig. 3. The overall pooled sensitivity of VM+ tumor was 0.82 (95% CI: 0.79-0.84, Fig. 4a), while the specificity of VM+ tumor 0.69 (95% CI: 0.66-0.71; Fig. 4b), among the 12 included studies. Furthermore, the overall pooled results for PLR, NLR, and DOR were 2.56 (95% CI: 1.94-3.93), 0.17 (95% CI: 0.07-0.42), and 17.75 (95% CI: 5.30-59.44), respectively.
Associations between VM+ and the possible demographics and clinicopathological features of MM patients are sorted in the Table 3. Table 3 reveals none of the above covariates contributed to the heterogeneity (all P-value > 0.05). Therefore, on the basis of those covariates, the pooled sensitivity, specificity, PLR, NLR, DOR, and AUC for significant sub-analysis parameters were measured. We found statistically significant associations between VM and sample size, VM and race, as well as between expression of VM and staining method of VM (Fig. 5). As shown in Fig. 5a and Table 3, VM+ is a potentially accurate prognostic biomarker in CD31-/PAS+ staining subgroup (P = 0.24, 95% CI: 0.15-0.35) than CD34-/PAS+ staining subgroup (P = 0.39, 95% CI: 0.27–0.42) and PAS+ staining subgroup (P = 0.40, 95% CI: 0.30–0.52). So, the CD31-/PAS+ staining methods are relatively accurate diagnostic methods for detection of the VM, with 75% sensitivity and 70% specificity. The subgroups analysis was conducted based on sample size (≤100 vs. >100; Fig. 5b). The proportion of population with a high sample size (3 studies with more than 100 MM cases) was 0.41 (95% CI: 0.28–0.56; P-value = 0.12); while that of a sample size with less than 100 MM patients (9 studies) was 0.31 (95% CI: 0.23-0.41; P-value < 0.001). Meanwhile, highest specificity, NLR, and AUC in sample size less than 100 suggested that VM is more accurate in less sample size diagnosis. Interestingly, our results show that the overexpression of the VM was a high risk prognosis factor in Asia populations (7 studies with 503 cases; P = 0.32; 95% CI: 0.23–0.42; P-value < 0.001; Fig. 5c). As seen in the Table 3 and Fig. 5C, the pooled sensitivity and specificity were higher in the Asian patients compared to Caucasian patients (85% vs. 69% and 78% vs. 68%, respectively). Moreover, we could not find any significant correlation between the VM+ melanoma samples with gender, age, Clark level, and location of sampling (Data not shown).
Table 3 Subgroup analyses of the included studies
Publication bias and sensitivity analysis
The publication bias and sensitivity were analyzed using Funnel plots and empirically utilizing regression tests according to Begg’s test. The analysis was carried out by precluding a single study at a time. A symmetric inverted funnel shape in this study shows from a ‘well-behaved’ data set, in which publication bias is unlikely. After the ten studies were excluded, there was no obvious statistical evidence for publication bias in our meta-analysis (t = 1.41; P-value = 0.19) (Fig. 6). Hence, the results of the current meta-analysis were stable and credible, due no noticeable publications bias inﬂuencing overall results.