Clinicopathological and Prognostic Value of Ki-67 Expression in Bladder Cancer: A Systematic Review and Meta-Analysis

Background Ki-67 is an established marker of cell proliferation, and the Ki-67 index correlates with the clinical course of several cancer types, including bladder cancer (BC). However, the clinicopathological and prognostic significance of Ki-67 in bladder cancer remains unclear. Therefore, we performed a systematic review and meta-analysis to clarify this relationship. Methods A comprehensive literature search for relevant studies published up to February 1, 2016, was performed using PubMed, Cochrane Library, Embase and ISI Web of Knowledge. The effects of Ki-67 expression on survival outcome in patients with BC and BC subtypes were evaluated. Furthermore, the relationship between Ki-67 expression and the clinicopathological features of BC were assessed. Results Thirty-one studies with 5147 bladder cancer patients were selected for evaluation. Ki-67 expression was significantly associated with shorter recurrence-free (HR 1.69, 95% CI: 1.33–2.14), progression-free (HR 1.89, 95% CI: 1.43–2.51), overall (HR 2.03, 95% CI: 1.31–3.16), and cancer-specific (HR 1.69, 95% CI: 1.47–1.95) survival. Moreover, whereas high expression was more common in high tumor stage, recurrence status, tumor size, there was no correlation between high Ki-67 expression and age, gender, smoking habits, and tumor number. Importantly, analysis of the different subgroups of BC suggested that significant correlations between high Ki-67 expression and survival outcome (recurrence-free/progression-free/overall/cancer-specific survival) are present only in European-American patients. Conclusion The present results indicate that over-expression of Ki-67 is distinctly correlated with poor patient survival. Ki-67 may serve as a valuable biomarker for prognosis in BC patients, particularly in non-Asian BC patients. The results suggest no significant association between Ki-67 expression and BC prognosis in Asian patients. Further efforts are needed to fully clarify this relationship.

tissues; (2) evaluation of the relationship between Ki-67 expression and BC clinicopathological parameters and prognosis; and (3) sufficient information to estimate the hazard ratio (HR) of recurrence-free survival (RFS), progression-free survival (PFS), overall survival (OS), and cancer-specific survival (CSS) and a 95% confidence intervals (CIs). Papers containing any of the following were excluded: (1) duplicate literature or duplicate data presented at conferences; (2) reviews, no available data, or abstract only; (3) studies of cancer cell lines and animal models; and (4) insufficient data to obtain HR and its standard error. For overlapping articles, only the highest-quality and most-recent literature were retained.

Data extraction and methodological assessment
The following information was recorded for each study: the first author's name, publication year, sample source, number of cases, median or mean of patient age, gender, cancer stage, antibody source and dilution, percentage rate of expression, and follow-up period. We preferred to collect multivariate analysis data. If data were not available, data from univariate analyses of survival outcomes were extracted. All data were extracted by two independent observers (ZMM and ZHC). The quality of the selected articles was assessed according to the Newcastle-Ottawa Scale (NOS) criteria [17]. If data could not be obtained from the literature, we regarded the related data as not available.

Statistical analysis
The statistical analysis was conducted using Review Manager 5.3 (Cochrane Collaboration, Oxford, UK) and STATA 14.0 (Stata Corporation, TX). HRs and 95% CIs were used to evaluate the relationships between Ki-67 expression and RFS, PFS, OS, and CSS rates. ORs (odds ratios) and 95% CIs were used to estimate the relationships between Ki-67 expression and clinicopathological parameters, including age, sex status, tumor stage, recurrence status, tumor number, and tumor size. The statistical significance of the pooled ORs and HRs was evaluated by the Z test. Heterogeneity among the studies was evaluated with Cochran's Q test and I 2 tests [18]. When the I 2 statistic results were 0-50%, a fixed-effect model was used to calculate parameters. If the I 2 statistic results were 50-100%, a random-effects model was considered more appropriate than a fixed-effects model. A p value < 0.05 was considered statistically significant. Funnel plots and Begg's test were used to evaluate potential publication bias [19].

Study characteristics
Our search strategy initially identified 412 articles. Following deduplication (n = 60), the two reviewers independently screened the identified titles and abstracts. After manually screening the titles and abstracts, 22 studies were excluded because they were case reports (n = 2), review articles (n = 6), conference abstracts (n = 4), meta-analysis (n = 2) or studies irrelevant to the human studies (n = 8). Seven articles were ultimately excluded due to overlap with previously reported studies (n = 4). Thus, 31 articles published from 2001 to 2016 were included in the final meta-analysis   (Fig 1).
The main characteristics of the 31 studies included in our meta-analysis are presented in S1 Table. Of the 31 studies, 5 were conducted in America, five in Germany, five in China, three in Greece, three in Spain, three in Korea, two in Italy, two in Japan, and one each in Portugal, Switzerland, and the UK. In 5 of the 31 studies, patients received intravesical BCG therapy. The follow-up period of the studies ranged from 2 months to 124 months. The age of the patients ranged from 21 to 97 years, and the overall proportion of males was 80.33%.

Relationships between Ki-67 expression and RFS in bladder cancer using different cut-off values
Subgroup analysis demonstrated that the relationship between Ki-67 expression and RFS was not significant using different Ki-67 cut-off values (10%, 25%, 50%). The pooled HRs and 95% CIs were as follows: 1. 56 Table).

Relationships between Ki-67 expression and clinicopathological parameters
In this meta-analysis, the relationships between clinicopathological characteristics such as age, gender, smoking habits, tumor stage, recurrence status, tumor number, and tumor size and elevated Ki-67 expression were compared on the basis of 31 studies. The results of the meta-analysis revealed significant associations between high Ki-67 expression and higher tumor stage (Ta vs. T1; Ta/1 vs. T2-4), recurrence status, and larger tumor size. The combined ORs and 95% CIs were as follows: OR 0.29, 95% CI 0. 19  Ki-67 Expression in Bladder Cancer

Publication bias
Publication bias was conducted by Begg's test for RFS and PFS of bladder carcinoma, with P values of 0.964 and 0.152, respectively. (Fig 4A and 4C). Quantitative assessment by Egger's test for RFS and PFS suggested that our analyses were stable (P = 0.350, P = 0.195) (Fig 4B and 4D).

Discussion
Increasing evidence indicates that BC genomes exhibiting the most complex alterations are associated with a high Ki-67 proliferation index [51]. Pichu et al. [52] reported that in BC cells, prior exposure to anti-Ki-67 siRNA induces tumor cells to undergo curcumin-induced growth arrest and apoptosis by non-p53 and non-p21-dependent signaling pathways, which may be useful for gene therapy. Wang et al. [53] reported that the combined effects of TP53 and Ki-67 revealed predictive value for NMIBC recurrence. However, the relationship between Ki-67 and outcome remains unclear, and the roles and clinical significance of Ki-67 expression in BC have not been thoroughly investigated [54].
In the present study, the analyses of the pooled data indicated that (1) BC patients with high Ki-67 expression had a lower survival rate; (2) high Ki-67 expression was associated with the more aggressive clinical stage and larger tumor size in BC patients; (3) aberrant Ki-67 expression was higher in recurrent BC than in non-recurrent BC; (4) Ki-67 expression was not strongly associated with age, gender, and tumor number in BC patients; (5) a strong relationship between poor prognostic indicators and Ki-67 expression was established only for European-American patients. The correlation between Ki-67 expression and survival outcome (RFS/PFS/OS/CSS) did not reach statistical significance in Asian patients. Our study provides insights on the results of individual studies focused on the hypothesis that Ki-67 is a prognostic factor for BC and suggests that adjuvant therapy may be helpful in the high-risk subgroup of patients. Although further validation and investigation are needed, these data provide new insights on the biological aggressiveness of BC in Asian versus in non-Asian patients.
The biological mechanism of Ki-67 explains its prognostic significance in BC. Ki-67 is an index of cell proliferation and a measure of cell growth fraction during the G1, S, G2 and M stages of the cell cycle and is widely applied in immunohistochemistry (IHC) to estimate the activities of cell proliferation in many cancers. Some researches investigated the relationships between the Ki-67 and distant metastases [55,56]. They found that Ki-67 expression was upregulated in the transforming growth factor-β1 (TGF-β1) treated tumors, and TGF-β1 promotes EMT (epithelial-to-mesenchymal transition), migration, and invasion in bladder cancer cells [57]. Furthermore, it was showed that highly Ki-67 may induce EMT by increasing the expression of vimentin, which enhances cancer cell invasion and metastatic [58]. The present meta-analysis is the first study to systematically evaluate the associations between Ki-67 expression and clinicopathological features and prognostic factors in BC. Ki-67 can be considered an oncogene, and its activation may contribute to tumor progression and poor prognosis. Based on this meta-analysis, we suggest that Ki-67 expression in BC tends to indicate a poor prognosis.
Several limitations of this study must be acknowledged. In the included studies, the antibodies used to detect Ki-67 expression were not identical (anti-Ki67 mAb and anti-MIB-1 mAb). The definitions of the cut-off value also differed. Clinical factors such as race, age, and the use of different chemotherapies in each study may also be sources of bias. Non-English studies, unpublished studies, and studies that did not provide sufficient data to calculate HRs were not included in the assessment of the predictive value of Ki-67 for survival. These approaches may have produced errors due to the inclusion of inaccurate readings. Finally, although we included 31 studies comprising 5147 cases in this meta-analysis, few studies were categorized for subgroup analysis, and several survival subgroup analyses data lack. Therefore, more welldesigned and large-scale trials are needed to confirm these findings.
In conclusion, our meta-analysis confirmed the significant associations between Ki-67 expression and clinicopathological features and prognostic factors in BC. Although subgroup analysis indicated no significant association between Ki-67 expression and BC prognosis in Asian patients. Our meta-analysis demonstrates that Ki-67 has a detrimental effect on clinicopathological features and recurrence status in BC. Therefore, Ki-67 could serve as an independent prognostic factor of RFS, PFS, OS and CSS in European-American patients. Ki-67 may be a novel candidate for BC genotyping and an indicator for predicting the prognosis of BC patients.