MMP-8 C-799 T, Lys460Thr, and Lys87Glu variants are not related to risk of cancer

Comprehensive literature search on PubMed, Web of Science, EMBASE (Excerpta Medica Database), and SinoMed (China Wanfang Databases) was carried out to identify all eligible case-control studies, prior to June 2019. Valid keyword search strings are as follows: (MMP-8 OR matrix metalloproteinases 8) AND (polymorphism OR variant OR mutation OR SNP) AND (carcinoma OR tumor OR malignancy OR cancer). Furthermore, we independently retrieved the references in the identified articles to screen other available studies, with no language restriction. For studies with overlapping data, we selected the most recently published ones.

Two authors (LFZ and YYM) independently completed the data extraction based on the selection criteria. Any potential disagreement was discussed comprehensively to obtain a final consensus. The main features of included studies are summarized, which includes: first author’s name, year of publication, origin and race, type of cancer, age range, total number of participants, genotyping assay of MMP-8 C-799 T, Lys460Thr, and Lys87Glu variants in cases and controls, P-value of Hardy-Weinberg equilibrium (HWE) in controls. The quality score of the eligible studies was evaluated by Newcastle-Ottawa Scale (NOS). The research was regarded as high-quality if it acquired six or more stars.

We calculated the OR with 95% CI to investigate the strength of the relationship between MMP-8 C-799 T, Lys460Thr, and Lys87Glu polymorphisms and cancer susceptibility. A total of five genetic models were adopted: allelic contrast (M-allele vs. W-allele, for C-799 T, T vs. C; for Lys460Thr, C vs. A; for Lys87Glu, A vs. G), homozygote model (MM vs. WW, for C-799 T, TT vs. CC; for Lys460Thr, CC vs. AA; for Lys87Glu, AA vs. GG), heterozygote model (MW vs. WW, for C-799 T, TC vs. CC; for Lys460Thr, CA vs. AA; for Lys87Glu, AG vs. GG), dominant comparison (MM + MW vs. WW, for C-799 T, TT + TC vs. CC; for Lys460Thr, CC+ CA vs. AA; for Lys87Glu, AA + AG vs. GG), recessive comparison (MM vs. MW + WW, for C-799 T, TT vs. TC + CC; for Lys460Thr, CC vs. CA + AA; for Lys87Glu, AA vs. AG + GG). Q-test was utilized to estimate the heterogeneity among enrolled researches. If the heterogeneity was absent (P > 0.05), the fixed-effects model was employed [36]; alternatively, the random-effects model was performed [37]. Stratified analyses were carried out according to race (Asian, Caucasian, and Latin), type of cancer (bladder cancer and other cancers), and source of control. Hardy-Weinberg equilibrium (HWE) in the control group was calculated using a Chi-squared test. Begg’s funnel plot and the Egger’s test were both performed to measure the possible publication bias. P values of Begg’s and Egger’s test more than 0.05 indicated the absence of publication bias. The STATA software (Version 11.0, Stata Corporation, College Station, TX, USA) was adopted for all the above analyses.

To further investigate whether the expression of MMP-8 has an impact on tumorigenesis, we employed the online TCGA database to evaluate the MMP-8 expression in bladder cancer tissue and control counterparts. The effect of the expression of MMP-8 on bladder cancer patients’ overall survival time was also assessed. Furthermore, we adopted bioinformatics tools, like Polyphen-2 (http://​genetics.​bwh.​harvard.​edu/​pph2/​), to predict the role of MMP-8 SNPs at the protein level.

Main characteristics of the eligible articles as well as the genotyping assay results of MMP-8 C-799 T, Lys460Thr, and Lys87Glu variants have been summarized in Table 1. A total of 13 publications containing 19 case-control studies for the MMP-8 polymorphisms in compliance with the inclusion criteria were finally identified in the present analysis. All eligible studies had NOS score more than 6. There were 4372 cancer patients and 5066 control participants for the analysis of MMP-8 C-799 T variant. Five articles were acquired for assessing the association of MMP-8 Lys460Thr variant on cancer susceptibility, including 3019 cases and 3544 control subjects. There were 749 cases and 1919 controls on the Lys87Glu polymorphism. The MAFs (minor allele frequencies) of MMP-8 C-799 T variants were shown in Fig. 1: African, 0.185; East Asian, 0.423; European, 0.411; South Asian, 0.350; and American, 0.440. For MMP-8 Lys460Thr polymorphism: African, 0.222; East Asian, 0.009; European, 0.044; South Asian, 0.060; and American, 0.060. The MAFs for Lys87Glu variant were: African, 0.290; East Asian, 0.432; European, 0.453; South Asian, 0.400; and American, 0.480. In stratified analysis by race, 13 case-control studies were conducted on Asian descendants; five were based on the Caucasian population and one was based on Latin descendants. In stratified analysis by cancer type, four studies concerned bladder cancer. The rest were focused on other cancers, such as lung cancer, hepatocellular carcinoma, malignant melanoma, oral cancer, ovary cancer, and gastric cancer. In stratified analysis by the source of control, 12 studies were population-based controls, and the rest seven were hospital-based studies.

Summarized results and details of the present analyses for the three MMP-8 polymorphisms and cancer risk are provided in Table 2. Overall analysis indicated that the relationship between the three MMP-8 variants and cancer susceptibility was not significant. The MMP-8 C-799 T variant is not associated with the susceptibility of cancer under all genetic models (allele contrast: OR = 0.98, 95% CI = 0.92–1.04, P_{heterogeneity} = 0.068, P = 0.429; TT vs. CC: OR = 0.94, 95% CI = 0.82–1.07, P_{heterogeneity} = 0.097, P = 0.362; heterozygote comparison: OR = 1.00, 95% CI = 0.92–1.09, P_{heterogeneity} = 0.193, P = 0.992; TT + TC vs. CC: OR = 0.98, 95% CI = 0.90–1.07, P_{heterogeneity} = 0.086, P = 0.666; recessive model: OR = 0.94, 95% CI = 0.83–1.07, P_{heterogeneity} = 0.249, P = 0.348). In subgroup analysis by cancer type, we also indicated no relationship between MMP-8 C-799 T variant and bladder cancer (T vs. C: OR = 0.97, 95% CI = 0.83–1.12, P = 0.671, Fig. 2; TT vs. CC: OR = 0.85, 95% CI = 0.61–1.17, P = 0.316; TC vs. CC: OR = 1.08, 95% CI = 0.87–1.33, P = 0.492; TT + TC vs. CC: OR = 1.02, 95% CI = 0.84–1.25, P = 0.813; TT vs. TC + CC: OR = 0.83, 95% CI = 0.61–1.12, P = 0.219) or other cancers (T vs. C: OR = 0.99, 95% CI = 0.89–1.11, P = 0.898; TT vs. CC: OR = 0.96, 95% CI = 0.83–1.11, P = 0.578; TC vs. CC: OR = 0.99, 95% CI = 0.89–1.09, P = 0.763; TT + TC vs. CC: OR = 1.00, 95% CI = 0.86–1.16, P = 0.965; TT vs. TC + CC: OR = 0.97, 95% CI = 0.84–1.11, P = 0.631). In stratified analysis by race and source of control, no significant association between this polymorphism and cancer susceptibility was demonstrated (Fig. 3). For the MMP-8 Lys460Thr variant, we also indicated no major association of this variant on cancer risk (C vs. A: OR = 0.94, 95% CI = 0.67–1.32, P_{heterogeneity} = 0.905, P = 0.729; CC vs. AA: OR = 0.83, 95% CI = 0.36–1.93, P_{heterogeneity} = 0.859, P = 0.669; CA vs. AA: OR = 1.00, 95% CI = 0.66–1.50, P_{heterogeneity} = 0.904, P = 0.994; TT + TC vs. CC: OR = 0.96, 95% CI = 0.67–1.40, P_{heterogeneity} = 0.886, P = 0.848; recessive model: OR = 0.83, 95% CI = 0.36–1.93, P_{heterogeneity} = 0.866, P = 0.669, Table 2). In addition, similar results were revealed for the association between the MMP-8 Lys87Glu variant and cancer risk in allelic contrast (OR = 1.05, 95% CI = 0.93–1.18, P value for heterogeneity = 0.968, P = 0.430); homozygote model (OR = 1.11, 95% CI = 0.88–1.42, P_{heterogeneity} = 0.953, P = 0.377); heterozygote comparison (OR = 0.96, 95% CI = 0.78–1.18, P value for heterogeneity = 0.647, P = 0.722); dominant model (OR = 1.01, 95% CI = 0.83–1.22, P_{heterogeneity} = 0.863, P = 0.928), and recessive comparison (OR = 1.13, 95% CI = 0.93–1.38, P_{heterogeneity} = 0.604, P = 0.222).

Results from the TCGA database, containing 408 primary tumor and 19 normal samples, revealed that MMP-8 expression was elevated in bladder cancer tissue as compared to their control counterpart (P < 0.01, Fig. 4a). Furthermore, we investigated whether the MMP-8 expression had an effect on the overall survival time of bladder carcinoma participants. However, neither higher MMP-8 expression group nor lower expression group would have an impact on the patients’ overall survival time (P < 0.05, Fig. 4b, c). In addition, we adopted the Polyphen-2 bioinformatics tool to analyze the associations between MMP-8 Lys460Thr (K460 T, rs35866072), and Lys87Glu (K87E, rs1940475) variants and protein damage. Mutations of these SNPs are predicted to be “BENIGN” with a score less than 0.05, which indicated that neither Lys460Thr nor Lys87Glu SNP may probably damage the protein of MMP-8 (Fig. 5).

Both Egger’s and Begg’s funnel plot were employed for appraisal of the publication bias when evaluating MMP-8 C-799 T, Lys460Thr, and Lys87Glu variants. No evidence of publication bias was acquired for MMP-8 C-799 T polymorphism (T-allele vs. C-allele, t = 0.37, P = 0.722; TT versus CC, t = 0.26, P = 0.801; TC vs. CC, t = 0.59, P = 0.567; TT + TC versus CC, t = 0.62, P = 0.552; TT versus TC + CC, t = 0.07, P = 0.945). For MMP-8 Lys460Thr variant: M-allele vs. W-allele, t = − 0.35, P = 0.752; MW vs. WW, t = − 0.71, P = 0.527; MM + MW vs. WW, t = − 0.68, P = 0.545. For Lys87Glu variant: A vs. G, t = 0.38, P = 0.771; AA vs. GG, t < − 0.01, P = 0.998; AG vs. GG, t = − 0.04, P = 0.975; AA + AG vs. GG, t = − 0.13, P = 0.916; AA vs. AG + GG, t = 1.71, P = 0.338. Outlines of the funnel plots were relatively symmetrical for overall cancer risk, implying no significant publication bias (Fig. 6).