Prognostic Value of Tissue Vascular Endothelial Growth Factor Expression in Bladder Cancer: a Meta-analysis

Objective: The prognostic role of vascular endothelial growth factor (VEGF) in bladder cancer remains controversial. This meta-analysis aimed to explore any association between overexpression and survival outcomes. Methods: We systematically searched for studies investigating the relationships between VEGF expression and outcome of bladder cancer patients. Study quality was assessed using the Newcastle-Ottawa Scale. After careful review, survival data were extracted from eligible studies. A meta-analysis was performed to generate combined hazard ratios (HRs) for overall survival (OS), disease-free survival (DFS) and disease-specific survival (DSS). Results: A total of 1,285 patients from 11 studies were included in the analysis. Our results showed that tissue VEGF overexpression in patients with bladder cancer was associated with poor prognosis in terms of OS (HR, 1.843; 95% CI, 1.231-2.759; P = 0.003), DFS (HR, 1.498; 95% CI, 1.255-1.787; P = 0.000) and DSS (HR, 1.562; 95% CI, 0.996-1.00; P = 0.052), though the difference for DSS was not statistically significant. In addition, there was no evidence of publication bias as suggested by Begg’s and Egger’s tests except for DFS (Begg’s test, P = 0.221; Egger’s test, P = 0.018). Conclusion: The present meta-analysis indicated elevated VEGF expression to be associated with a poor prognosis in patients with bladder cancer.


Introduction
Bladder cancer is the second most common malignancy of the urinary tract after prostate cancer, with approximately 390 000 new cases annually, and has the sixth highest cancer mortality (Jemal et al., 2011).Despite recent advances in screening and multimodality therapy, the outcome for bladder cancer remains generally poor, emphasizing the need for early detection and prognostic markers.Currently, the most widely studied prognostic factors are related to pathological characteristics of the neoplasm, including tumor size, grade, stage, and vascular invasion (Thieblemont et al., 1996;Kanda et al., 2006;Youssef et al., 2011;van Rhijn, 2012).However, a variety of other potential prognostic markers remain to be further characterized (Kanda et al., 2006).
Angiogenesis, defined as the formation of new blood vessels from existing vasculature, plays an important role in tumor growth and metastasis by providing oxygen, nutrients and growth factors to the cancer cells (Folkman,
Therefore, in this study, we sought to conduct a metaanalysis to estimate the prognostic importance of elevated VEGF expression for survival among patients with bladder cancer.

Search strategy
We searched Medline, PubMed, Embase, and the Web of Science using the search terms: ((VEGF or vascular endothelial growth factor) and (cancer or carcinoma) and 'bladder' and 'prognosis').The last search was updated in November 2012.To expand our search, references of the retrieved articles were also screened for additional studies.

Study selection
Two independent reviewers assessed the eligibility of studies.Abstracts of all candidate articles were read.Articles that could not be categorized based on title and abstract alone were retrieved for full-text review.These articles were independently read and checked for inclusion criteria.Disagreements were resolved through consensus with a third reviewer.
Primary studies that reported data required for metaanalysis were identified and included.Study inclusion/ exclusion criteria criteria for primary studies were as follows: (1) proven diagnosis of bladder cancer in humans, (2) VEGF evaluation using tissue-based methods, and (3) correlation of VEGF with overall survival (OS), diseasefree survival (DFS) or disease-specific survival (DSS).There was no pre-specified sample size or follow-up period used to determine study inclusion.Only studies written in English were included.Studies not directly reporting hazard ratios (HRs) were allowed if data were available for statistical estimation.When more than one of the same patient populations was included in several publications, only the most recent or complete study was used to avoid duplication of information.

Quality assessment
Quality assessment was performed in each of the acceptable studies by two reviewers independently using the Newcastle-Ottawa Quality Assessment Scale for cohort studies (Table 1) (Wells et al., 2003).This scale is an eight-item instrument that allows for assessment of patient population and selection, study comparability, follow-up, and outcome of interest.A star system of the NOS (range, 0-9 stars) has been developed for the evaluation.The highest value for quality assessment was 9 stars.Any discrepancies were resolved by a consensus reviewer.

Data extraction
Two investigators extracted data from eligible studies independently, discussed discrepancies and reached consensus for all items.The following information was extracted from each article: (1) basic information from papers, such as first author's name, year of publication; (2) information of study designation, such as study design; (3) demographic data such as inclusion criteria, patient age, sex, and treatment during follow-up; (4) tumor data such as VEGF expression in the primary site, stage, grade, vascular invasion, and metastases; (5) survival data such as OS, DFS and DSS; (6) variables such as number of patients analyzed, method of tissue VEGF measurement, cut-off values for VEGF levels, and geographical district of the patients.The primary data were the HRs and 95% confidence intervals (CsI) for survival outcomes, including OS, DFS and DSS.

Statistical analysis
The primary outcome for analysis was survival in patients with high VEGF values as compared to those with low VEGF values.HRs with 95% CIs were reported for individual studies with HR>1 and 95% CI for the aggregated HR not crossing 1 designates a prognostic role of high VEGF.When HRs were not reported in an article, they were calculated to use established methods reported by Parmar et al. (1998).
Forrest plots were undertaken to evaluate the heterogeneity of combined HRs.Statistical assessment was performed using a χ 2 -based test of homogeneity and evaluation of the inconsistency index (I 2 ) statistic.Heterogeneity was defined as p<0.10 or I 2 >50% (Higgins et al., 2003).When heterogeneity was judged between primary studies, a fixed effect model was used for pooled analyses.If not, a random effect model was used (DerSimonian et al., 1986).Egger's test was performed to test for publication bias (Egger et al., 1997).
Statistically significant test was determined by a P-value of less than 0.05 for a summary HR and publication bias.All analyses were carried out using STATA.Summary table of studies included in meta-analysis.Study design is described as case-controlled (C) or retrospective (R).Treatment describes whether the patients received curative surgical resection (S).Tumour grade was most often described using the WHO classification, but occasionally other systems were utilized.Study quality is listed using the results of the Newcastle -Ottawa questionnaire (Table 1); NR, not reported

Study identification and eligibility
An electronic search yielded 118 articles, of which 71 were excluded on the basis of their abstracts.We then screened the remaining 47 articles in full text.Upon further review, 17 articles was eliminated on the basis of without survival data, 11 articles were excluded because there is no special result of VEGF, and 6 articles were eliminated due to inadequate data for calculation.We also excluded a previous study with data overlap and a study investigating the association of serum VEGF level with survival.The selection process and reasons for exclusion have been summarized in Figure 1.From the 11 studies that were included (Bernardini et al., 2001;Theodoropoulos et al., 2004;Yang et al., 2004;Suzuki et al., 2005;Zu et al., 2006;Herrmann et al., 2007;Nadaoka et al., 2008;Szarvas et al., 2008;Pignot et al., 2009;Shariat et al., 2010;Li et al., 2011;Zaravinos et al., 2012), a total of 1285 patients were analyzed.The characteristics of the selected studies are presented in Table 2.

Quality assessment
Quality assessment using the Newcastle-Ottawa Scale was performed on all 11 studies included for meta-analysis.Of note, there was no study attempting to control for important prognostic factors that may have confounded the association of high VEGF with survival.The NOS scores of 1-3, 4-6 and 7-9 were defined as low, intermediate and high-quality studies, respectively.Our NOS results showed that the median overall score was 5 (range 4 to 7), which indicated that the quality of included trials was acceptable.

Disease-specific survival
The pooled hazard ratio didn't show significant difference in DFS between VEGF over expression group with VEGF low expression group 1.562; 95% CI, 0.996-1.00;P = 0.052; Figure 4).There was significant heterogeneity (P = 0.032, I 2 =59.2, χ 2 =12.24), and the pooled HR for DFS was performed by using the randomeffects model.

Publication bias
There was no evidence for significant publication bias in OS (Begg's test, P = 0.133; Egger test, P = 0.149) and DSS (Begg's test, P =1; Egger test, P = 0.140) studies.However, according to DFS, Begg's test indicated no publication bias among these studies regarding risk ratio (P = 0.221), but Egger's test indicated a publication bias (P = 0.018).

Discussion
angiogenesis is one of the hallmarks of cancer (Hanahan et al., 2011) , and VEGF, as one of the most important regulators in tumor angiogenesis (Ferrara, 2004), has been thought to be valuable of predicting poor outcome of survival in several cancers, such as lung cancer (Zhan et al., 2009), hepatocellular carcinoma (Schoenleber et al., 2009), gastric cancer (Chen et al., 2011), ovarian cancer (Yu et al., 2012), and osteosarcoma (Qu et al., 2012).However, the prognostic value of VEGF in bladder cancer was undetermined.Yang et al. (2004) and Theodoropoulos et al. (2004) both announce the association between VEGF over expression and poor outcome of patients with bladder cancer.Then several following studies supported their results (Zu et al., 2006;Pignot et al., 2009;Shariat et al., 2010;Li et al., 2011).However, several other studies demonstrated that the level of VEGF expression did not predict outcomes for patients with bladder cancer (Suzuki et al., 2005;Nadaoka et al., 2008;Szarvas et al., 2008;Zaravinos et al., 2012).As a result, the role of VEGF expression in bladder cancer is still not well defined.
There were several potential sources of heterogeneity among the studies.First, Studies might differ in the characteristics of included patients (age, histological type, tumor grade, stage, tumor size, treatment received, and the duration of follow-up).Furthermore, in some studies, patients were excluded because of insufficient tissue source, insufficient clinical data or insufficient survival data.All of these could potentially lead to selection bias or recruitment bias.Second, language also induces a bias, as positive results tend to be published in English in international journals.Although our search was not restricted, all the studies included were written in English.Third, the differences of methodology and cut-off values among included studies also were sources of heterogeneity and caused selection biases potentially.The variability of IHC techniques, which we could not avoided, may prevent tissue VEGF measurements from standardization.Although four studies chose the median VEGF level as the cut-off value, values varied among studies obviously.Additionally, the heterogeneity in tissue samples cannot be ignored.Fourth, observers in some studies were not blinded to the outcome data, which contributed to information bias.
Publication bias is another problem that we should consider in the present meta-analysis.In order to minimize publication bias, we did the literature search as completely as possible, using PubMed and EMBASE databases, screening references of the retrieved articles, and looking over posters from the annual meetings of the European Society of Medical Oncology and the American Society of Medical Oncology.However, missing some data was unavoidable.In our study, we did not adopt abstracts because data were not available in abstracts.In addition, positive results have more tendencies to be accepted by journals, other than negative results.What's more, negative results are often not submitted for review by journals.Therefore, publication bias was still detected for DFS.
Several important limitations need to be considered when interpreting our analysis.First and important of all, significant heterogeneity between studies existed indeed.Second, this meta-analysis relied on published trials rather than individual patient data (IPD), and meta-analyses based on published data tend to overestimate the predictive effects of VEGF compared with individual patient data analyses.In addition, it precludes a more comprehensive analysis such as adjusting for baseline factors and other differences that existed between the trials from which the data were pooled.Third, original studies included in our analyses almost were retrospective studies (10/11) with 1  doi.org/10.7314/APJCP.2013.14.2.645 Prognostic Value of Tissue VEGF in Bladder Cancer: a Meta-analysis case-controlled study, providing a lower level of evidence.And furthermore, publication bias was even we expanded our search to the best of our ability.Finally, the small number of included patients negated the possibility of exploring possible sub-group analyses and explored heterogeneity among study populations.
In the conclusion, this meta-analysis for the first time demonstrated that high levels of VEGF are associated with a poor prognosis in patients with bladder cancer.However, one should be cautious when interrupting these results due to the limitations of our studies.Further high-quality studies are still needed to confirm these results.

Selection 1 .
Representativeness of the exposed cohort a) Truly representative of the average 'bladder cancer patient' in the community (*) b) Somewhat representative of the average 'bladder cancer patient' in the community (*) c) Selected group of users (e.g.nurses, volunteers) d) No description of the derivation of the cohort 2. Selection of the non exposed cohort a) Drawn from the same community as the exposed cohort (*) b) Drawn from a different source c) No description of the derivation of the non exposed cohort 3. Ascertainment of exposure a) Secure record (e.g.surgical records) (*) b) Structured interview (*) c) Written self-report d) No description 4. Demonstration that outcome of interest was not present at start of study a) Yes (*) b) No Comparability 1. Comparability of cohorts on the basis of the design or analysis a) Study controls for 'metastasis or micrometastasis' (*) b) Study controls for any additional factor (*) Outcome 1. Assessment of outcome a) Independent blind assessment (*) b) Record linkage (*) c) Self-report d) No description 2. Was follow-up long enough for outcomes to occur?a) Yes ('2 years') (*) b) No 3. Adequacy of follow-up of cohorts a) Complete follow-up -all subjects accounted for (*) b) Subjects lost to follow-up unlikely to introduce bias -small number lost '25%' follow up, or description of those lost (*) c) Follow up rate < '75%' and no description of those lost d) No statement A study can be awarded a maximum of one star (*) for each numbered item in the Selection and Outcome categories.A maximum of two stars can be given in Comparability.Underlined and quoted phrases are provided in the scale to allow for adjustment to particular studies DOI:http://dx.doi.org/10.7314/APJCP.2013.14.2.645  Prognostic Value of Tissue VEGF in Bladder Cancer: a Meta-

Figure
Figure 1.Flow Chart of the Meta-analysis

Figure 4 .
Figure 4. Random-effects Model of Hazard Ratio (95% confidence interval) of DSS Associated with High VEGF Levels Versus Low Levels