Introduction
Cancer remains one of the leading causes of mortality worldwide, with an estimated 20 million new cases and 9.7 million cancer-related mortalities globally in 2022 (1,2). The incidence and mortality rates vary across different types of cancer, among which breast, lung, colorectal, prostate and gastric cancer are the most common malignancies (3–5). Lung cancer continues to be the primary cause of cancer-related mortality, accounting for ~18.7% of all cancer-associated mortalities globally (1). Genetic factors play a key role in the occurrence and progression of cancer. In recent years, numerous genetic polymorphisms have been shown to be associated with susceptibility to various types of cancer (6,7).
Matrix metalloproteinases (MMPs) are a family of zinc-dependent proteases that can degrade the extracellular matrix (ECM), playing notable roles in tumor growth, invasion and metastasis (8,9). MMPs, particularly MMP-1, MMP-3 and MMP-8, are closely involved in several key processes of tumorigenesis, including apoptosis, angiogenesis and immune evasion (10–12). MMP-1 is widely expressed, participates in the degradation of type I, II and III collagen, and promotes tumor cell proliferation and metastasis by remodeling the tumor microenvironment (13). As an MMP, MMP-3 can hydrolyze various ECM components, such as laminin and collagen, thereby facilitating tumor cell migration and invasion (14–17). MMP-8, functioning as a collagenase, may affect the carcinogenic process by altering the balance of the tumor microenvironment (18,19).
Although numerous studies have investigated the association between MMP polymorphisms and cancer susceptibility, the results have shown considerable heterogeneity. For example, the MMP-1-1607 polymorphism (rs1799750), a 1G/2G insertion/deletion variant, has been associated with an increased risk of bladder, colorectal and esophageal cancer, but with a reduced risk of prostate cancer in certain populations (20–22). Similarly, inconsistent findings have also been reported for MMP-3 and MMP-8 polymorphisms. A systematic review summarized that the MMP-8 rs11225395 polymorphism was associated with reduced risks of breast and bladder cancer, but increased risks of melanoma and ovarian cancer, while rs2155052 was associated with decreased lung cancer risk (18). By contrast, an updated meta-analysis found no significant association between the MMP-3-1171 5A/6A polymorphism and lung cancer risk (23). These differences in study outcomes may be attributed to variations in study design, population characteristics and the genetic models employed.
Given these inconsistencies, the present study conducted a systematic review and meta-analysis to evaluate the relationships between MMP-1, MMP-3 and MMP-8 polymorphisms and cancer risk. The primary aim of the present study was to clarify whether these polymorphisms are universally associated with cancer susceptibility across different populations and cancer types, and to assess the functional significance of these genetic variants by utilizing publicly available genomic data, thereby providing a comprehensive understanding of their roles in cancer risk.
Discussion
MMP-1, MMP-3 and MMP-8 polymorphisms have been associated with tumor susceptibility, but results have been inconsistent. The present meta-analysis, based on data from 36,368 patients and 40,246 controls, assessed the associations between these polymorphisms and tumor susceptibility. A total of three polymorphisms were significantly associated with cancer risk: MMP-3 rs35068180 and rs3025058, and MMP-1 rs1799750. Using the Venice criteria and FPRP test, the cumulative evidence showed four strong associations (MMP-3 rs3025058 with esophageal cancer in the overall population under the dominant model, MMP-1 rs1799750 with glioblastoma in the overall population under the recessive model, and MMP-1 rs1799750 with renal cancer in the overall population under both the recessive and allelic models), six moderate associations and 14 weak associations.
MMP-1 (NCBI Gene ID: 4312; http://www.ncbi.nlm.nih.gov/gene/) is expressed in tumor cells and predicts poor survival (for example, in colorectal, gastric, pancreatic, lung, breast, prostate, bladder and oral/head-and-neck cancer) (38). The MMP-1 rs1799750 (1G>2G) variant, located in the promoter region, enhances MMP-1 transcription by creating an Ets-binding site. The 2G allele is associated with active matrix degradation and tumor progression (39). Increasing evidence has linked MMP-1 rs1799750 (1G>2G) to several types of cancer, including bladder, colorectal and gastric cancer (40–42). The meta-analysis performed in the present study found that this SNP significantly increases glioblastoma risk by 2.036-fold in the overall population under the recessive model (985 samples), with the association upgraded to strong (FPRP<0.05). Studies suggest that members of the E26 transformation-specific family of transcription factors in glioma cells bind to the 2G promoter, possibly contributing to glioblastoma (43,44). Additionally, MMP-1 rs1799750 was shown to be associated with increased renal cell carcinoma risk under both allelic and recessive models (1,569 samples), with a 1.351-fold increased risk under the allelic model and a 1.674-fold increased risk under the recessive model. However, no subgroup analysis by ethnicity was possible, likely due to the small sample size. Further studies, particularly in different ethnicity groups, are warranted.
MMP-3 (Gene ID: 4314) located within the MMP cluster on chromosome 11q22.3, is expressed by epithelial cells, fibroblasts and macrophages, contributing to ECM remodeling. Through broad ECM-degrading activity and activation of other pro-MMPs, MMP-3 promotes tumor invasion and progression (10,45,46). The MMP-3 promoter contains a 5A/6A polymorphism at −1612, with the 5A allele showing twice the promoter activity of the 6A allele in in vitro assays, suggesting that transcription inhibitors may bind to the 6A allele (47). The present study found a strong association between MMP-3 rs3025058 and esophageal cancer risk under the dominant model (1,620 samples), with the 5A allele increasing risk by 1.563-fold (OR, 1.563; 95% CI, 1.165–2.04). This association was observed in European populations but is limited by the lack of data on Asian populations, which warrants further investigation.
In the present study, six associations with cancer risk were rated as moderate evidence. A total of three associations (colorectal cancer with MMP-1 rs1799750, esophageal cancer with MMP-1 rs1799750 and glioblastoma with MMP-1 rs1799750) were upgraded to strong evidence, graded ‘ACA’, ‘ABC’ and ‘AAC’, respectively, by the Venice criteria due to replication and small-study effects. With FPRP<0.05, these were considered medium-strength associations with tumor susceptibility. In total, two associations in Asian individuals (colorectal cancer with MMP-1 rs1799750 and MMP-3 rs35068180) were downgraded from strong to moderate (FPRP>0.2). The remaining associations (colorectal cancer with MMP-1 rs1799750) were not upgraded or downgraded (0.05≤ FPRP ≤0.20), with the relation rated ‘BAA’ by the Venice criteria due to the amount of evidence. Further analysis revealed that these moderate-evidence associations exhibited considerable complexity across different populations, cancer types and genetic models. A possible mechanism is that polymorphisms in genes such as MMP-1 and MMP-3 may influence transcription factor binding, promoter activity or linked functional variants, thereby affecting MMP expression and activation in a tissue- and environment-dependent manner and contributing to cancer susceptibility (48,49). Tumor heterogeneity, diverse genetic backgrounds among populations, environmental factors, as well as differences in sample size and methodological design, may all contribute to the observed variability (50,51). Therefore, the differential effects of the same SNP in various populations and cancer types warrant further systematic investigation in future studies.
In the present study, 14 associations were classified as weak regarding tumor susceptibility. MMP-1 rs1799750 was significantly associated with several types of cancer (colorectal cancer, gastric cancer, nasopharyngeal carcinoma, bladder cancer, esophageal cancer and glioblastoma), although the association with colorectal cancer in Caucasian individuals under the allelic model was rated as weak (FPRP=0.488), despite moderate heterogeneity (‘ABA’ by Venice criteria). Sample size, environmental factors and methodology may explain variations across ethnicities. A total of two associations (gastric cancer with MMP-1 rs1799750 and glioblastoma with MMP-1 rs1799750) were also rated as weak, mainly due to low ORs and HWE deviations; thus, further studies are needed. MMP-3 rs3025058 (recessive/allelic models) showed a weak association with esophageal cancer, with Venice criteria ratings ‘ACC’ and ‘BCC’ due to limited study populations. No significant association was found between MMP-8 polymorphism and cancer risk, warranting larger sample sizes for confirmation.
Although the present study identified several statistically significant associations, the majority of effect sizes were modest, indicating that individual MMP variants (including rs1799750) are unlikely to provide sufficient predictive value as stand-alone clinical markers. For example, even for the strong association between MMP-1 rs1799750 and glioblastoma, the pooled effect (OR=2.036 under the recessive model) reflects relative risk rather than absolute risk and thus does not directly translate into high positive predictive value (PPV) in the general population. Instead, the most plausible translational pathway is risk stratification in high-risk settings, where germline variants may help refine risk when combined with other determinants. In future work, rs1799750 and other credible MMP loci could be incorporated into multi-locus genetic scores or polygenic risk models and further integrated with established environmental/clinical factors (for example, smoking, chronic inflammation, hormonal status or occupational exposures) to improve discrimination and calibration. Prospective, multi-ethnic cohort studies that include baseline incidence and exposure data will be required to estimate absolute risk, PPV and model performance (for example, area under the curve) and to determine whether adding MMP variants meaningfully enhances existing prevention or screening frameworks (52–56).
Overall, the present findings are broadly consistent with previous meta-analyses, although the strength and direction of some associations appear to vary across cancer types and populations (57–60). The present study adds notable value by incorporating a larger sample size, applying more rigorous Venice criteria and FPRP evaluation, and conducting more detailed subgroup analyses by ethnicity and tumor subtype. Notably, the present study identified context-dependent and ethnicity-specific associations that were previously unrecognized or only partially addressed. Previous studies have shown that MMP-1 rs1799750, MMP-3 rs3025058 and other MMP variants may influence cancer susceptibility, albeit with moderate effect sizes and considerable heterogeneity across populations and tumor types (48,61). The present study further refines these associations by applying rigorous Venice criteria and FPRP testing, supporting strong epidemiological credibility for only a small subset of associations (notably, MMP-1 rs1799750 with glioblastoma and renal cancer, as well as MMP-3 rs3025058 with esophageal cancer). Mechanistically, MMP-1 and MMP-3 variants may modulate gene expression by altering promoter or enhancer regions, thus affecting the degradation of ECM components, tumor cell invasion, angiogenesis and immune cell infiltration in the tumor microenvironment (45). Functional annotation from ENCODE and eQTL data performed in the present study suggests that rs1799750 and rs3025058 could influence transcription factor binding and gene expression across multiple tissues, further supporting their biological plausibility in tumorigenesis. However, the overall moderate or weak evidence for the majority of associations in the present meta-analysis also suggests that MMP variants alone are unlikely to be major determinants of cancer risk and may interact with environmental or lifestyle factors such as smoking, inflammation or hormonal status.
In the present study, associations varied across ethnicities, genetic models and tumor types. While the majority of studies focused on Asian populations, research on non-Asian populations is key. Different genetic models showed varying associations with tumors, and gaps in research on certain tumor types highlight the need for larger, more focused studies.
The present study revealed that four SNPs in three MMPs were not associated with seven types of cancer in any genetic model and/or ethnicity (Table III). Among these, two SNPs in two MMPs showed no association with the risk of two types of cancer in meta-analyses, including a minimum of 2,000 cases and 2,000 controls, which offers >85% power to detect an OR of 1.15 under the additive model for a variant with minor allelic frequency 0.20, Type 1 error 0.05. Therefore, further research on these SNPs with a similar sample size may not yield positive results.
Several limitations of the present study should be acknowledged. First, the meta-analysis included studies only published in English, potentially introducing language and publication bias. Second, the present analysis was primarily restricted to case-control studies, limiting causal inference. In addition, the selection of hospital-vs. population-based controls varied across studies, potentially introducing selection bias, and survival bias may exist if genotype affects both cancer incidence and prognosis. Third, subgroup analyses could not account for all possible confounders, such as age, sex, environmental exposures or comorbidities, due to limited available data. Furthermore, the observed heterogeneity (I2≤81.7%) may reflect differences in population genetic structure, analytical models, clinical characteristics of included cases and genotyping platforms across studies. These factors are well-recognized sources of variability in genetic epidemiology and likely contributed to the inconsistent effect estimates. Another limitation is that some ethnicity-stratified analyses were based on a limited number of studies (for example, the association between MMP-1 rs1799750 and renal cancer), which restricts the robustness of population-specific conclusions and warrants caution when generalizing these findings across diverse ancestral groups. Additionally, the broad ethnicity classifications (such as Asian and Caucasian) of the present study may inadequately capture population substructure and admixture patterns within these groups, as different subpopulations may exhibit distinct allele frequencies and LD structures. Fourth, only a subset of functionally annotated SNPs was investigated and gene-gene or gene-environment interactions were not systematically evaluated. Moreover, since individual-level baseline risk and exposure data were unavailable in the included studies, the present study could not estimate absolute risk measures or predictive metrics such as PPV, negative PV or area under the curve for any single SNP. It is also possible that the analyzed SNPs may represent LD proxies rather than causal variants themselves and population-specific differences in LD patterns could contribute to observed heterogeneity. Technical heterogeneity across studies, including different genotyping platforms and quality control thresholds, represents a potential source of systematic bias. To address this and minimize bias, a strict sensitivity analysis based on HWE was performed. Since deviations from HWE often indicate genotyping errors, this analysis acted as a rigorous quality filter. The results demonstrated that the primary findings, specifically the strong associations between MMP-1 rs1799750 and glioblastoma and renal cancer, remained robust after this strict exclusion. This resilience suggests that the core conclusions of the present study are not driven by technical variations or platform-specific biases, whereas less robust associations (which lost significance) should be interpreted with caution. Additionally, the exclusive focus of the present study on common variants overlooks potentially important contributions from rare variants, which may exhibit larger effect sizes and account for additional heritability in cancer susceptibility. Fifth, the present study focused only on cancer susceptibility, without addressing prognosis, response to therapy or functional effects at the cellular level. Similarly, the present study treated broad cancer types as homogeneous entities without considering histological subtypes, molecular classifications or disease stages that may have distinct genetic architectures. Finally, temporal changes in diagnostic criteria and treatment protocols across the present study period spanning multiple decades may affect case definition consistency.
Future research should include larger, multi-ethnic cohorts, more comprehensive genotyping, and integrated analyses of environmental risk factors, lifestyle and somatic mutations. Experimental studies investigating the functional impact of MMP variants on protein expression, enzyme activity and tumor biology are also warranted. Furthermore, emerging genomic and transcriptomic data from single-cell and spatial profiling technologies may shed light on the context-dependent effects of MMP polymorphisms in different tumor microenvironments (62). The present study identified four strong associations between gene mutations and cancer risk (MMP-3 rs3025058 in esophageal cancer under the dominant model, MMP-1 rs1799750 in glioblastoma under the recessive model, and in renal cancer under both allelic and recessive models across the overall population) as well as six moderate associations and 14 weak associations. These findings may help to identify populations at increased risk of cancer and provide a comprehensive summary of current evidence regarding MMP polymorphisms and cancer risk, thereby offering valuable insights for future studies. Overall, the present comprehensive meta-analysis highlights the complex and context-dependent associations between MMP-1, MMP-3 and MMP-8 gene variants and cancer risk, emphasizing the importance of multi-dimensional approaches in genetic epidemiology and laying the groundwork for further exploration of the interplay between ECM remodeling, genetic susceptibility and cancer development.