Although the matrix metalloproteinase-1 (MMP1) polymorphism MMP1–1607 (1G>2G) has been associated with susceptibility to various cancers, these findings are controversial. Therefore, we conducted this meta-analysis to explore the association between MMP1–1607 (1G>2G) and cancer risk. A systematic search of literature through PubMed, Embase, ISI Web of Knowledge, and Google Scholar yielded 77 articles with 21,327 cancer patients and 23,245 controls. The association between the MMP1–1607 (1G>2G) polymorphism and cancer risks was detected in an allele model (2G vs. 1G, overall risk [OR]: 1.174, 95% confidence interval [CI]: 1.107–1.244), a dominant model (2G2G/1G2G vs. 1G1G OR, OR: 1.192, 95% CI: 1.090–1.303), and a recessive model (2G2G vs. 1G2G/1G1G, OR: 1.231, 95% CI: 1.141–1.329). In subgroup analysis, these associations were detected in both Asians and Caucasians. After stratification by cancer types, associations were found in lung, colorectal, nervous system, renal, bladder, and nasopharyngeal cancers. This meta-analysis revealed that MMP1–1607 (1G>2G) polymorphism was significantly associated with elevated risk of cancers.

1. Introduction

Single-nucleotide polymorphisms (SNP) are variations in single nucleotides that occur at specific positions in the genome and influence protein structure, gene splicing, transcription factor binding, messenger RNA degradation, or sequences of noncoding RNAs [1]. SNPs reportedly contribute to interindividual variability in susceptibility to common diseases such as cancer.

Matrix metalloproteinases (MMPs) are a group of proteolytic enzymes that can degrade extracellular matrix components, thereby affecting various physiological and pathological processes such as embryonic development, wound healing, arthritis, atherosclerosis, and tumor progression [2]. Increasing evidence shows that MMPs play significant roles in cancer development, including cell growth, differentiation, apoptosis, angiogenesis, invasion, and metastasis [3].

MMP1, a member of the MMP family, can degrade interstitial collagen types I, II, and III, clearing a path for cancer cells to invade matrix barriers and migrate through tissue stroma [4]. The MMP1 gene is located at 11q22.3, and MMP1 expression can be regulated by the MMP1 promoter. The gene polymorphism MMP1–1607 (1G>2G) or rs1799750 in the MMP1 promoter has been associated with increased susceptibility for various cancers [5, 6]. However, the results were controversial because of variations in cancer types and patient demographics. Therefore, we conducted this meta-analysis to further explore the association between MMP1–1607 (1G>2G) polymorphism and cancer susceptibility.

2. Materials and Methods

2.1. Identification and Eligibility of Studies

We conducted a systematic search of literature published until December 2017 that investigated the association of MMP1–1607 (1G>2G) polymorphism with cancer risks, through PubMed, Embase, ISI Web of Knowledge, and Google Scholar, using the terms “Matrix metalloproteinase-1 or MMP-1 or rs1799750,” “polymorphism or variation or mutation or SNP,” and “cancer or carcinoma or tumor or neoplasm.” Only case–control studies with sufficient genotype distribution data to calculate odds ratios (ORs) with 95% confidence interval (CIs) in different gene models were included. Letters, case reports, animal studies, and reviews were excluded. When overlapping populations were included in different articles, only the publication with the largest sample size was selected.

2.2. Data Extraction

Two investigators independently reviewed the articles to exclude irrelevant and overlapping studies. The following data were extracted from eligible publications: first author, published year, cancer type, country, ethnicity, control source, genotyping method, and genotype distribution. Any disagreements were resolved by discussion or by consultation with another investigator.

2.3. Statistical Analysis

The meta-analysis was conducted using SATAT (version 13.0). The Hardy–Weinberg equilibrium (HWE) for control groups was checked by the chi-square goodness-of-fit test () The associations between MMP1–1607 (1G>2G) polymorphism and cancer risks were calculated by OR and 95% CI with the following models to avoid assuming only one suboptimal genetic model: an allele model (2G vs. 1G), a dominant model (2G2G/1G2G vs. 1G1G), and a recessive model (2G2G vs. 1G2G/1G1G). Subgroup analyses were performed by cancer type and ethnicity.

The heterogeneity of studies was assessed by test using value and value. A fixed-effects model was adopted when test indicated a lack of heterogeneity (); otherwise, a random-effects model was used. We considered 0–40% of value to indicate low heterogeneity, 30–60% to indicate moderate heterogeneity, 50–90% to indicate substantial heterogeneity, and 75–100% to indicate considerable heterogeneity. Publication bias was measured with funnel plots and Harbord’s and Peter’s tests.

3. Results

3.1. Characteristics of Eligible Studies

The study selection procedure is shown in Figure 1. We included 77 articles with 21,327 cancer patients and 23,245 controls in this meta-analysis (Table 1) [783]. Of these, 43 articles were conducted among Asian populations and 34 among Caucasian populations; 67 studies were hospital-based and 10 were population-based. Of the different genotyping methods used in these studies, 45 used polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP), 18 used TaqMan real-time PCR, 8 used sequencing, and 6 used other methods. Sixteen of the 77 articles showed deviations from HWE in control groups.

3.2. Quantitative Analysis

The main results of this meta-analysis are listed in Table 2. The association between the MMP1–1607 (1G>2G) polymorphism and cancer risks was seen in the allele model (2G vs. 1G, OR: 1.174, 95% CI: 1.107–1.244; Figure 2), the dominant model (2G2G/1G2G vs. 1G1G, OR: 1.192, 95% CI: 1.090–1.303; Figure 3), and the recessive model (2G2G vs. 1G2G/1G1G, OR: 1.231, 95% CI: 1.141–1.329; Figure 4).

3.3. Risk by Cancer Type

When we considered different cancer types, elevated risk was found in lung cancer in the allele model (2G vs. 1G, OR: 1.128, 95% CI: 1.002–1.268) and the dominant model (2G2G/1G2G vs. 1G1G, OR: 1.127, 95% CI: 1.005–1.264).

Significant association was also found in colorectal cancer in the allele model (2G vs. 1G, OR: 1.279, 95% CI: 1.087–1.505), the dominant model (2G2G/1G2G vs. 1G1G, OR: 1.281, 95% CI: 1.033–1.588), and the recessive model (2G2G vs. 1G2G/1G1G, OR: 1.368, 95% CI: 1.094–1.712).

Five articles addressed the MMP1–1607 polymorphism in nervous system cancers, including astrocytoma, glioblastoma, hypophyseal adenoma, and malignant gliomas. Significantly elevated risks were observed in all the three different models (2G vs. 1G, OR: 1.799, 95% CI: 1.493–2.168; 2G2G/1G2G vs. 1G1G, OR: 2.070, 95% CI: 1.474–2.906; and 2G2G vs. 1G2G/1G1G, OR: 1.935, 95% CI: 1.498–2.501).

In renal cancer, the association was found in the allele model (2G vs. 1G: OR: 1.351, 95% CI: 1.149–1.590) and the recessive model (2G2G vs. 1G2G/1G1G OR: 1.674, 95% CI: 1.351–2.073). In bladder cancer, only in the recessive model was significant association detected (2G2G vs. 1G2G/1G1G, OR: 1.739, 95% CI: 1.074–2.816).

Increased risk was also found in nasopharyngeal cancer in the allele model (2G vs. 1G, OR: 1.212, 95% CI: 1.067–1.377) and the recessive model (2G2G vs. 1G2G/1G1G, OR: 1.267, 95% CI: 1.074–1.488).

No relationship was observed in gastric cancer, oral cancer, ovarian cancer, breast cancer, prostate cancer, head and neck cancer, endometrial cancer, hepatocellular cancer, or esophageal cancer (Table 2).

3.4. Risk by Ethnicity

In the Asian population, the association between the variation and cancer risks was detected in the allele model (2G vs. 1G, OR: 1.228, 95% CI: 1.130–1.334), the dominant model (2G2G/1G2G vs. 1G1G, OR: 1.256, 95% CI: 1.084–1.456), and the recessive model (2G2G vs. 1G2G/1G1G, OR: 1.297, 95% CI: 1.176–1.431). In the Caucasian population, evaluated risk was also found in the allele model (2G vs. 1G, OR: 1.109, 95% CI: 1.023–1.202), the dominant model (2G2G/1G2G vs. 1G1G, OR: 1.126, 95% CI: 1.015–1.249), and the recessive model (2G2G vs. 1G2G/1G1G, OR: 1.431, 95% CI: 1.013–1.289). Although significant differences were observed in both Asian and Caucasian populations, the Asian population showed higher risk than the Caucasian for the allele, dominant model, or homozygous model, but showed a decreasing trend in the recessive model (Table 2).

3.5. Heterogeneity and Sensitivity Analysis

Heterogeneity was observed in overall analyses in all comparison models with and range from 50.2% to 74.0% (indicating moderate or substantial heterogeneity). We therefore used the random-effects model. Sensitivity analysis to assess influence of individual studies showed no individual study to greatly affect the pooled OR.

3.6. Publication Bias

The forest plot seemed to be symmetrical (Figure 5). Harbord’s and Peter’s tests revealed no statistical significance in publication bias (Harbord’s: ; Peter’s: ).

4. Discussion

The MMP1–1607 (1G>2G) polymorphism has been associated with increased transcription of MMP1 due to an insert of a guanine base that creates a core-binding site for the EST family of transcription factors, which leads to increased susceptibility for tumor occurrence and progress. The significant association between the variation of MMP1–1607 (1G>2G) with some cancer types has been reported by different meta-analyses [3, 4, 8486].

In the current meta-analysis of 77 articles with 21,327 cancer patients and 23,245 controls, the MMP1–1607 (1G>2G) polymorphism was a strong risk factor in various cancers. Although both Asian and Caucasian individuals with 2G alleles or 2G2G genotypes may be more susceptible to cancer development, several studies revealed significant associations in Asians, but not Caucasians [5, 6]. These discrepancies might be due to limited sample sizes. Moreover, the Asian population seemed to show increased risk compared with Caucasian populations when the allele or dominant models were adopted, whereas a decreasing trend was observed in a recessive model, which implies different susceptibilities.

The association was found in lung, colorectal, nervous system, renal, bladder, and nasopharyngeal cancers, but not gastric, oral, ovarian, breast, prostate, head-and-neck, endometrial, hepatocellular, or esophageal cancers, which indicates that the variation plays different roles in various cancers, in accordance with pervious meta-analyses [4, 85, 87, 88]. However, these papers only focused on single types of cancer or one specific ethnicity. Our meta-analysis included all the cancers, analyzed the overall pooled OR, and performed subgroup analyses. Our findings imply a complex relationship between cancer susceptibility and gene variation, influenced by cancer sites and ethnicities.

Recently, the functional studies of SNPs have moved fast. For instance, a study reported that a missense variant rs149418249 in the TPP1 gene confers colorectal cancer risk by interrupting TPP1–TIN2 interaction and influencing telomere length [89]. An expression quantitative trait locus-based analysis revealed that a mutation rs27437, residing in the upstream of SLC22A5, can affect colorectal cancer risk by regulating SLC22A5 expression [90]. Another article reported that a TCF7L2 missense variant rs138649767 associates with colorectal cancer risk by interacting with a GWAS-identified regulatory variant rs698326 in the MYC enhancer [91]. However, the biological mechanisms of functional SNPs still remain challenging. Therefore, further studies are required to promulgate the real functions by which the MMP1–1607 (1G>2G) polymorphism may influence cancer susceptibility and progression.

Our study had some limitations. First, moderate or substantial heterogeneity was detected between studies, which was not significantly decreased by subgroup analysis. When all variations were included in the meta-regression analysis, no obvious factors were detected. More subgroup analyses should be performed, based on factors such as tobacco or alcohol consumption. This conclusion should be interpreted with caution. Second, this analysis was performed with candidate gene strategy in which the MMP1–1607 (1G>2G) polymorphism was selected for study based on a priori knowledge of the gene’s biological functional impact on the trait or disease in question [92]. Genome-wide association studies (GWAS) which scan the entire genome for genetic variation include immense amounts of SNPs. Published papers usually reported those SNPs with highly statistical significance (usually ). We have retrieved literature through PubMed in order to search the evidence of association between the MMP1–1607 (1G>2G) polymorphism and cancer risks in GWAS results [92, 93]. However, we did not acquire any positive findings. We speculate that ethnic discrepancy, population stratification, and different standards of statistical significance might lead to negative findings in GWAS. Third, due to the innate shortage of case–control designed studies, the quantity of studies was limited. Third, gene–gene and gene–environment interactions should be considered in analyses of the effects of genes. Fourth, more original papers with large sample sizes were required due to lack of eligible studies in specific cancers in this analysis.

5. Conclusions

In conclusion, an association between the MMP1–1607 (1G>2G) polymorphism and cancer risks was detected in both Asians and Caucasians. After stratification by cancer types, associations were found for lung cancer, colorectal cancer, nervous system cancer, renal cancer, bladder cancer, and nasopharyngeal cancer. More original studies with larger sample size are required for future analysis.

Conflicts of Interest

The authors declare no competing financial interests.


This study was partly funded by the National Natural Science Foundation of China (Grant No. NSFC 81502195 and NSFC 81672512) and Medicine and Health Science Technology Development Project of Shandong Province (No. 2016WS0258). We thank Liwen Bianji, Edanz Group China (http://www.liwenbianji.cn/ac), for editing the English text of a draft of this manuscript.