BackgroundMatrix metalloproteinases (MMPs) are regarded to be relevant to the prognosis of breast cancer. Numerous studies have confirmed the association between MMPs and tumor growth, invasion and metastasis in breast cancer. However, their prognostic values for survival in patients with breast cancer remain controversial. Hence, a meta-analysis was performed to clarify a more accurate estimation of the role of MMPs on prognosis of breast cancer patients.MethodA systemic electronic search was conducted in PubMed, Embase and Web of science databases to identify eligible studies, which were associated with the relationship between MMPs and prognosis of breast cancer. The correlation in random-effect model was evaluated by using the hazard ratios (HRs) and 95% confidence intervals (CIs).ResultsA total of 28 studies covering 4944 patients were included for meta-analysis. A summary hazard ratio (HR) of all studies was calculated, as well as the sub-group HRs. The combined HRs calculated by either univariate or multivariate analysis both suggested that overexpression of MMPs had an unfavorable impact on overall survival (OS) (HR = 1.694, 95%CI: 1.347–2.129, P < 0.001; HR = 1.611, 95%CI: 1.419–1.830, P < 0.001, respectively). And the univariate analysis showed that patients with overexpression of MMPs had worse relapse-free survival (RFS) (HR = 1.969, 95%CI: 1.460–2.655, P < 0.001) in all eligible studies. In the sub-group analyses, HRs of MMP-9 positivity with poor OS were 1.794 (95%CI: 1.330–2.420, P < 0.001) and 1.709 (95%CI: 1.157–2.526, P = 0.007) which were separately evaluated by univariate and multivariate analysis. A small number of articles demonstrated that MMP-2 overexpression was not related with shorter OS (HR = 1.400, 95%CI: 0.610–3.029, P = 0.427). Four studies included in the OS analysis of MMPs expression in serum suggested that positive expression of serum MMPs may be an unfavorable factor (HR = 1.630, 95%CI: 1.065–2.494) for breast cancer patients. No publication bias was observed in the current meta-analysis.ConclusionsOur findings suggested that MMPs overexpression (especially MMP-9, MMP-2, MMPs overexpression in serum) might indicate a higher risk of poor prognosis in breast cancer. Larger prospective studies are further needed to estimate the prognostic values of MMPs overexpression.