Clinicopathologic significance and prognostic value of Ki-67 expression in patients with gastric cancer: a meta-analysis

Background The prognostic value and clinicopathologic significance of Ki-67 expression in gastric cancer patients was controversial. This meta-analysis was performed to clarify the prognostic value and clinicopathologic significance of Ki-67 expression in gastric cancer patients. Materials and Methods Several electronic databases were searched for eligible studies. The pooled odds ratio (OR), hazard ratios (HR) and 95% confidence interval(CI) were calculated to explore the prognostic value and clinicopathologic significance of Ki-67 expression for disease free survival and overall survival. Results Totally 5600 gastric cancer patients from 29 studies were included in this study. High Ki-67 expression was significantly related with Lauren's classification (OR = 1.70; P = 0.001; 95%CI: 1.40-2.06) and tumor size(OR = 1.54; P = 0.006; 95%CI: 1.14-2.09). However, high Ki-67 expression was not significantly associated with lymph node metastasis (OR = 1.37; P = 0.138; 95% CI: 0.90-2.08), tumor stage (OR = 1.31; P = 0.296; 95% CI: 0.79-2.16) and tumor differentiation (OR = 1.03; P = 0.839; 95% CI: 0.78-1.35). The pooled HRs were 1.87(P = 0.001; 95% CI 1.30-2.69) for disease free survival and 1.23(P = 0.005; 95% CI 1.06-1.42) for overall survival. Conclusions High Ki-67 expression may serve as a predictive biomarker for poor prognosis in gastric cancer patients. Stratification by Ki-67 expression may be a consideration for selection of therapeutic regimen and integrated managements.


INTRODUCTION
Gastric cancer (GC) is the fourth malignant tumor and the second leading cause of tumor related death in the world [1]. Patients with advanced GC has only a median overall survival (OS) of less than 12 months [2][3]. Therefore, there is an urgent need for reliable prognostic factors to predict poor prognosis and to subdivide different risk stratification for management of GC patients.
Ki-67 is a nuclear protein which expresses throughout the cell cycle in proliferating cells [4]. The correlation between Ki-67 expression and prognosis of GC patients were still contradictory . Meanwhile, the clinicopathologic significance of Ki-67 expression in GC patients was uncertain. Therefore, we performed this meta-analysis to determine the clinicopathologic significance and prognostic value of Ki-67 expression in GC patients. Review www.impactjournals.com/oncotarget

Search results
The initial search returned 595 articles (with 75duplicate articles). After screening the abstracts, 445 irrelevant articles were excluded. Reviewers identified 75 potential studies for full-text review and 46 articles were eliminated due to inadequate data. Finally, 29 studies were included in the present study . The details of screening process were shown in Figure 1.

Study selection and characteristics
The characteristic of included studies were summarized in Table 1. The publication time of included studies was between 1996 and 2016.The study sample size was between 56 and 693, with a mean sample size of 193. The NOS score of 29 studies varied from 7 to 8, with a mean value of 7.73. Twenty-three studies provided survival information and fifteen studies presented clinicopathologic parameters.

Prognostic value of high Ki-67 expression in gastric cancer patients
A total of 4741 GC patients from 23 eligible studies were included and analyzed for prognostic value of Ki-67 expression in GC patients( Figure 3 and Figure 4). The pooled HRs was 1.87(P = 0.001; 95% CI 1.30-2.69) for DFS and 1.23(P = 0.005; 95% CI 1.06-1.42) for OS.

Sensitivity analysis
All studies were sequentially removed to explore that whether any individual study had an significant influence to the pooled HR. The pooled HR in sensitivity meta-analysis ranged from 1.17(95%CI: 1.02-1.34) to 1.28 (95%CI: 1.11-1.48) for OS, demonstrating that the pooled HR was not significantly affected by any individual study (Table 2).   Stability assessment of the pooled hazard ratios of Ki-67 expression for overall survival by cumulative meta-analysis The pooled HRs of cumulative meta-analysis( Figure  6)

Explore of sources of heterogeneity by metaregression analyses and subgroup analyses
The pooled HR was 1.31(95% CI 1.05-1.62; P = 0.017; heterogeneity = 82.7%; P < 0.001) in studies with patient number more than 100 for OS whereas it was 1.08(95% CI 0.91-1.28; P = 0.386; heterogeneity = 26.3%; P < 0.228) in studies with patient number not more than 100 (Table 3). The results suggested that sample size might contributed to the clinical heterogeneity. However, meta-regression analysis did not find any source of heterogeneity (all P > 0.05, data not shown).  Some previous studies have reported that high Ki-67 expression was associated with poor OS in GC patients [14,16,22,24,26,32]. These original studies have revealed that high Ki-67 expression had a predictive value for prognosis of GC patients. Our conclusion was consistent with that of these previous studies. Recently, several meta analyses have reported that high Ki-67 expression was associated with poor prognosis in different tumors, including gastrointestinal stromal tumor, cervical cancer and non-small cell lung cancer [35][36][37]. Furthermore, two studies have further reported that Ki-67 expression could be used for risk stratification in patients with gastrointestinal stromal tumor [38][39].
The heterogeneity was significant in the present meta-analysis. There might be some potential sources of heterogeneity as follows: First, the heterogeneity caused by different cut-off values of Ki-67 expression was inevitable. Second, subgroup analyses showed that sample size might be a potential source of heterogeneity. Third, the identify methods of Ki-67 expression (TMAS or WTS) and the number of count cells(1000 or 500) might yield variation in different studies. In addition, heterogeneity could be caused by other factors, such as study regions, pathology types, tumor stages, treatments and races. Although significant heterogeneity existed in the present meta-analysis, sensitivity analyses and cumulative meta-analyses demonstrated that HRs of Ki-67 expression for prognosis of GC patients was stable and reliable. Furthermore, we performed Begg's funnel plot and Egger's test to assess the potential publication bias and did not find any evidence of publication bias.
The present meta analysis had several strengths: Firstly, we first explored the association between Ki-67 expression and clinicopathologic parameters in GC  patients. Secondly, we included 29 eligible studies and 5600 patients, which could strengthen persuasiveness of the conclusions. Thirdly, Ki-67 expression in 29 eligible studies was all detected by IHC. Fourthly, studies published in Chinese were included as English literature to increase representation of study population.
The results of the present meta analysis need to be interpreted cautiously for several limitations. First, most studies defined positive status of Ki-67 expression according to different cut-off values. Second, heterogeneity was inevitable due to different baseline characteristics. Third, although the method for extracting survival information from survival curve is widely accepted, we could not completely eliminate the sources of information inaccuracy in the process of extracting data.
In conclusion, high Ki-67 expression may serve as a predictive biomarker for poor prognosis in gastric cancer patients. Stratification by Ki-67 expression may be a consideration for selection of therapeutic regimen and integrated managements.

Search strategy
Several electronic databases, including PubMed, EMBASE, Cochrane Library, Web of Knowledge, China National Knowledge Infrastructure and WanFang data, were searched from January 1970 to May 2016. We performed literature search by combined text word and MeSH(Emtree for EMBASE database accordingly) strategy with terms " Ki-67 Antigen" or "MIB-1 Antigen" and "gastric cancer" or "gastric carcinoma" or "stomach tumor" and "survival" or "outcome" or "prognosis" or "prognostic". The strategy was correspondingly adjusted in different databases. In the retrieval process, expanded search of hyponym was performed. We made a manual search using the reference lists of the relevant articles. We contacted the corresponding author to get necessary information if necessary. The search was restricted to human studies, but there was no restriction on language or publication time. All clinical investigation and data achievement were conducted according to the principles expressed in the Declaration of Helsinki.

Criteria for inclusion and exclusion
The inclusion criteria were as follows: (1) proven pathological diagnosis of GC in humans; (2) Ki-67 expression evaluation using immunohistochemistry (IHC) method; (3)provided information on clinicopathological parameters and/or overall survival information. Studies not directly providing survival information were included if survival information were available from survival curve. Articles published in Chinese were included as English literature. Only the most recent study was included among duplicate studies. There were no restrictions on sample size or follow-up period.
The following studies were excluded: (1) reviews, letters, case reports, and conference abstracts without original data; (2) non-human experiments;(3) laboratory studies;(4)articles from which the necessary information could not be extracted.

Quality assessment of studies
Two reviewers (Zhiqiao Zhang and Jinxin Lin) independently assessed the quality of studies using the Newcastle-Ottawa Quality Assessment Scale(NOS) ( Table  1). Disagreements were resolved through consensus with a third reviewer (Guanying Luo).

Data extraction
Two investigators (Zhiqiao Zhang and Jinxin Lin) independently extracted and examined the following data: surname of the first author, publication year, country, sample size, disease stage, detection method of Ki-67, clinical parameters and survival outcome data. Study information was extracted and recorded using a standardized form. All eligible studies were coded as surname of the first author + publish year in the standardized form. Study authors were contacted to obtain key information if necessary. When necessary, a third investigator (Guanying Luo) helped to reach a consensus.

Statistical analysis
The statistical analysis was performed according to the guidelines suggested by the Meta-Analysis of Observational Studies in Epidemiology group(MOOSE) [34]. The pooled odds ratio (OR) were combined to explore the association between KI-67 expression and clinicopathological parameters. The pooled hazard ratio (HR) were used to summary outcome of overall survival. While survival data were not directly reported, we extracted survival information from Kaplan-Meier curve. The heterogeneity among different studies was measured by the Q and I 2 tests. A probability value of I 2 ≥30% and P < 0.1 indicated the existence of significant heterogeneity. A random effect model (DerSimonian and Laird method) or fixed effect model(Mantel-Haenszel method) was used depending on the results of heterogeneity analysis. The potential publication bias was assessed by Begg's funnel plot and Egger's test. P value < 0.05 was considered statistically significant. The statistical analyses were performed by STATA version 12.0 software (Stata Corporation,College Station, Texas, USA).

Abbreviations
Gastric cancer GC odds ratio OR hazard ratio HR confidence interval CI.

Author contribution
Peng Wang,and Zhiqiao Zhang designed the study. Zhiqiao Zhang and Jixin Lin performed the research; Zhiqiao Zhang and Jixin Lin collected and analyzed the data; Zhiqiao Zhang and Jixin Lin wrote the paper; Yunzhao Hu, Guanying Luo, Zhaowen Luo , and Canchang Cheng amended the article. Zhiqiao Zhang act as the submission's guarantor and takes responsibility for the integrity of the work as a whole, from inception to published article. All authors reviewed the manuscript.

ACKNOWLEDGMENTS
This study was funded in part by Health Department and Finance Department of Guangdong Province.

CONFLICTS OF INTERESTS
The authors have declared that they had no conflict of interests.