Association between matrix metalloproteinases polymorphisms and ovarian cancer risk: A meta-analysis and systematic review

Background Published data on the relationship between matrix metalloproteinases (MMPs) polymorphisms and ovarian cancer risk have implicated inconclusive results. To evaluate the role of MMPs polymorphisms in ovarian cancer risk, a meta-analysis and systematic review were performed. Methods MMPs polymorphisms which could be quantitatively synthesized were involved in meta-analysis. Five comparison models (homozygote model, heterozygote model, dominant model, recessive model, additive model) were carried out, a subgroup analysis was performed to clarify heterogeneity source. The remaining polymorphisms which could not be quantitatively synthesized were involved in systematic review. Results 10 articles with 20 studies were included in this paper. Among those studies, 8 studies involving MMP1 rs1799750 and MMP3 rs34093618 could be meta-analyzed and 12 studies involving 12 polymorphisms could not. Meta-analysis showed that no associations were found between MMP1 rs1799750 (homozygote model: OR = 0.93, 95%CI = 0.70–1.23, POR = 0.60; heterozygote model: OR = 1.09, 95%CI = 0.78–1.54, POR = 0.61; dominant model: OR = 1.02, 95%CI = 0.83–1.25, POR = 0.84; recessive model: OR = 0.95, 95%CI = 0.75–1.21, POR = 0.67; additive model: OR = 1.00, 95%CI = 0.85–1.17, POR = 0.99), MMP3 rs34093618 (homozygote model: OR = 1.25, 95%CI = 0.70–2.24, POR = 0.46; heterozygote model: OR = 1.08, 95%CI = 0.51–2.31, POR = 0.84; dominant model: OR = 0.97, 95%CI = 0.68–1.38, POR = 0.85; recessive model: OR = 1.12, 95%CI = 0.69–1.80, POR = 0.65; additive model: OR = 1.01, 95%CI = 0.79–1.31, POR = 0.91) and ovarian cancer. Furthermore, similar results were detected in subgroup analysis. The systematic review on 12 polymorphisms suggested that MMP2 C-735T, MMP7 A-181G, MMP8 rs11225395, MMP9 rs6094237, MMP12 rs2276109, MMP20 rs2292730, MMP20 rs12278250, MMP20 rs9787933 might have a potential effect on ovarian cancer risk. Conclusions In summary, polymorphisms of MMPs might not be associated with ovarian cancer risk. However, it is necessary to conduct more larger-scale, multicenter, and high-quality studies in the future.


Results
10 articles with 20 studies were included in this paper. Among those studies, 8 studies involving MMP1 rs1799750 and MMP3 rs34093618 could be meta-analyzed and 12 studies involving 12 polymorphisms could not. Meta-analysis showed that no associations were found between MMP1 rs1799750 (homozygote model: OR = 0.93, 95%CI = 0.70-1. 23

Introduction
Ovarian cancer is main cause of death with gynecological tumors worldwide, and is often at an advanced stage by the time of diagnosis and has metastasized throughout the peritoneal cavity [1][2]. In 2013, there were an estimated 22,240 new cases and 14,030 new deaths [3]. Despite continuous advances in ovarian cancer research, diagnosis, and clinical treatment during the past 30 years [4], it has been still hard to find a cost-effective screening strategy to significantly increase the survival rate for early-stage ovarian cancer. Genome-wide association studies (GWAS) concerning genetic aetiology of cancer have established more than 150 regions associated with various specific cancers, which expand the current understanding of carcinogenesis mechanisms [5]. Alterations in genetic sequence, such as single-nucleotide substitutions, lead to cancer formation by biologically regulating a handful of molecular activities [6].
Matrix metalloproteinases (MMPs), a family of more than 20 zinc-dependent enzymes known to degrade extracellular matrix and basement membrane components [7], are not only a prerequisite for multiple steps of cancer development but also play important roles in cancer invasion and metastasis [8]. MMPs are correlated with ovarian cancer, with the levels of MMP-2, MMP-7 and MMP-9 elevated in ovarian cancer patients [9][10]. At genetic level, a number of studies have been carried out to assess the association between polymorphisms of MMPs and ovarian cancer risk [11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27], but the conclusions have been still conflicted and even contradictory. For example, study by Ju [19] showed no associations existed between MMP1 rs1799750 and ovarian cancer in Korean, while study by Kanamori [11] showed 2G genotype of MMP1 rs1799750 might represent a risk factor for ovarian cancer in Japanese. Individual studies with a small sample size may result in incorrect conclusion. Therefore, a comprehensive meta-analysis and systematic review are necessary to precisely assess the relationships between MMPs polymorphisms and ovarian cancer risk.

Search strategy
The databases Pubmed, Embase, Web of knowledge, were searched for all articles with the following search terms: (MMP OR MMPs OR matrix metalloproteinase OR matrix metalloproteinases) AND (polymorphism OR polymorphisms) AND (ovarian cancer OR ovarian carcinoma) up to search date: March 25, 2017. No limitation of publication language was defined for this search. Additional published data were identified by reviewing the bibliographical references listed in each retrieved article.

Inclusion criteria and exclusion criteria
All studies included in this meta-analysis were accorded with the following inclusion criteria: (a) study focused on the association between MMPs polymorphisms and ovarian cancer; (b) case-control design; (c) provided available frequency for each genotype in both cases and controls to calculate odds ratio (OR) and corresponding 95% confidence interval (95%CI). In addition, exclusion criteria were as follows: (a) reviews, editorials, comments or animal studies; (b) overlapped articles or studies with overlapping data.

Data extraction
Two investigators independently extracted the following data: first author's name, year of publication, study country, ethnicity, source of controls, MMPs gene, polymorphisms, number of cases and controls, value of Hardy-Weinberg equilibrium (HWE). A consensus on the extracted items was reached by discussion between the two investigators.

Quality assessment
The quality of study was assessed according to the quality assessment criteria [28] (S1 Table), in which the quality scores ranged from 0 to 15. Studies with scores !9 were regarded as high quality.

Statistical analysis
In order to evaluate the association between MMPs polymorphisms and ovarian cancer risk, OR and 95% CI were summarized under five comparison models, including homozygote was checked by a chi-square-based Q statistic test and quantified by I 2 value. If I 2 value < 50% or P > 0.10, the fixed effect model was used [29]. Otherwise, random effect model was carried out [30], then a subgroup analysis by ethnicity was performed. Both funnel plot and Egger's test were performed to test whether publication bias existed or not, bias was considered with P<0.05 in Egger's test. The statistical analyses for the present study were completed by Review Manager software 5.1 (the Nordic Cochrane Center, Rigshospitalet, Copenhagen, Denmark) and Stata software 12.0 (StataCorp, College Station, TX, USA).

Heterogeneity analysis and subgroup analysis
For both MMP1 rs1799750 and MMP3 rs34093618 polymorphism, there was obvious heterogeneity in heterozygote model (MMP1 rs1799750: I 2 = 53%, P het = 0.08; MMP3 rs34093618: I 2 = 69%, P het = 0.04). Then, a subgroup analysis by ethnicity was conducted to assess the source of heterogeneity. The forest plots of subgroup analysis for MMP1 rs1799750 and MMP3 rs34093618 were respectively presented in Figs 12 and 13. For MMP1 rs1799750, heterogeneity dramatically decreased when stratification analyses for Caucasian was conducted (I 2 = 31%, P het = 0.15), while MMP3 rs34093618 did not decreased (I 2 = 81%, P het = 0.02). No significant association was found between MMPs polymorphism and ovarian cancer in both two subgroup analysis. Matrix metalloproteinases polymorphisms and ovarian cancer risk evidence of publication bias in the meta-analysis of MMP1 rs1799750 polymorphism. For MMP3 rs34093618, publication bias analysis was not conducted for only 3 studies involved.

Discussion
Study by Ju [19] showed no associations existed between MMP1 rs1799750 and ovarian cancer in Korean, while study by Kanamori [11] showed 2G genotype of MMP1 rs1799750 might represent a risk factor for ovarian cancer in Japanese. Therefore, a comprehensive meta-analysis and systematic review are necessary. As a powerful tool for summarizing the different studies, meta-analysis has been accepted as a significant tool to analyze cumulative data from limited study subjects [31].
This meta-analysis and systematic review, including 5 studies for MMP1 rs1799750 composed of 754 ovarian cancer cases 1184 and controls, 3 studies for MMP3 rs34093618 polymorphism composed of 265 cases and 425 controls, 12 studies for systematic review involving 2793 cases and 3037 controls, proved that MMP1 rs1799750 and MMP3 rs34093618 polymorphisms were not associated with ovarian cancer risk, in addition, subgroup analyses by ethnicity showed similar results. Although in systematic review eight polymorphisms, including MMP2 C-735T, MMP7 A-181G, MMP8 rs11225395, MMP9 rs6094237, MMP12 rs2276109, MMP20 rs2292730, MMP20 rs12278250, MMP20 rs9787933, might be associated with ovarian cancer risk, it was inconclusive results due to lack of relevant studies. Except eight above polymorphisms, it was revealed that other four polymorphisms in systematic review were not related with ovarian cancer risk.
The major strengths of our study were its comprehensive and systematic focus on the relationship between MMPs polymorphisms and ovarian cancer risk. Although a meta-analysis by Wang [32] has also investigated the relationship of MMP1 rs1799750 polymorphism with ovarian cancer (5 studies involving 754 cases and 1184 control) and produced similar results, our report identified 15 additional studies including 3058 cases and 3462 controls, which have not been included in report of Wang [32].
Also, some limitations still existed in our paper. First, control group was not uniformly defined, some controls were population-based while other controls were hospital-based. Second, significant heterogeneity was observed in a few comparison models. Although a subgroup analysis was performed to clarify sources, it was hard to find all potential sources. Third, departure from HWE was detected in some studies. Finally, there was a lack of a unified criterion for including studies, leading to failure to adjust them in age and lifestyle et al.
In summary, our reports showed that MMPs polymorphisms might not be associated with ovarian cancer risk. However, it is necessary to conduct more larger-scale, multicenter, and high-quality studies in the future.
Supporting information S1