La MMP2 est une protéinase stromale qui dégrade plusieurs éléments de la matrice extracellulaire et qui active d’autres MMPs dont la MMP9. La MMP14 active la MMP2 via l’inhibiteur de la pro-MMP2, le TIMP2. Cette étude avait pour buts de démontrer l’association entre la surexpression de la MMP2 et la récidive du CaP, d’évaluer l’influence de la MMP9, de la MMP14 et du TIMP2 sur cette association, puis de tester par un modèle murin l’influence de la MMP14 stromale sur la croissance tumorale. Nous avons analysé immunohistochimiquement 204 cas de CaP (T3NxM0) et avons étudié séparément les cellules cancéreuses, stromales et épithéliales bénignes (CEB), notamment selon le pourcentage d’expression des marqueurs (< 10% ou ≥ 10%). Le suivi médian est de 4,61 ans. La surexpression de la MMP9 dans les cellules cancéreuses est associée à un score de Gleason élevé (8-10) (p = 0,0009). Dans les CEB, la surexpression la MMP14 est associée au niveau de PSA sérique initial bas (≤ 20 ng/ml) (p = 0,0027). Le TIMP2 est protecteur (RR = 0,573, p = 0,0233) lorsque fortement exprimé dans les cellules stromales. La MMP2 augmente le risque de récidive lorsqu’exprimée par les cellules cancéreuses (RR = 1,549, p = 0,0027, test de tendance). L’étude de la croissance de tumeurs injectées à des souris nues/nues suggère que la MMP14 est associée à l’implantation tumorale. La MMP9 et la MMP14 seraient impliquées dans l’évolution du CaP alors que la MMP2 et le TIMP2 peuvent servir de marqueurs pronostiques pour les CaP T3NxM0. / Introduction: The matrix metalloproteinase 2 (MMP2) is associated with poor prognosis in many neoplasms. MMP14 activates MMP2 using pro-MMP2 specific inhibitor TIMP2 as a receptor. Activated MMP2 degrades extracellular matrix components such as collagen and gelatin, and activates other MMPs including MMP9 (gelatinase B). We therefore tested the influence of MMP9, MMP14 and TIMP2 expression on prostate cancer (Pca) disease-free survival and the association between MMP2, MMP9, MMP14 and TIMP2. Stromal MMP14 involvement in tumor growth was also tested. Material and methods: By immunohistochemistry, we analyzed 200 T3NxM0 Pca cases. We evaluated marker expression separately in cancer, stromal and benign epithelial (BE) cells according to a percentage scale (0, < 10, 10-50 and ≥ 50%) and to low (< 10%) or high (≥ 10%) expression of the markers. MCF7 cells were injected subcutaneously to nu/nu mice with MMP14 -/- fibroblasts and their growth was compared to MCF7 + MMP14 +/+ fibroblasts tumors. Results: Median follow-up was 4.61 years. MMP9 overexpression in cancer cells was associated with high Gleason score (8-10) (p = 0.0009). Low initial PSA serum levels (≤ 20ng/ml) were indirectly associated with MMP14 overexpression in BE cells (p = 0.0027) and with MMP9 overexpression in stromal (p = 0.0050) and BE cells (p = 0.0056). There was a decreased risk of Pca recurrence with high (≥ 10%) TIMP2 expression in stromal cells (hazard ratio (HR) = 0.573, p = 0.0233) and an increased risk of Pca recurrence with MMP2 expression by > 50% of BE cells (HR = 3.006, p = 0.0387). Increased risk of Pca recurrence was also observed with high MMP2 expression (HR = 1.549, p = 0.0027, trend test) and with the following combinations: low TIMP2 in stromal cells and high MMP2 in BE cells (HR = 4.121, p < 0.0001); high MMP2 in stromal cells and low TIMP2 in cancer cells (HR = 2.742, p = 0.0171). MCF7 studies suggest MMP14 stromal involvement in Pca implantation. Conclusions: MMP9 and MMP14 are involved mostly in Pca implantation. MMP2 and TIMP2 might be used as predictors of disease-free survival in T3NxM0 Pca.