Plasminogen activator inhibitor-I as a measure of vascular remodelling in breast cancer

The generation of urokinase plasminogen activator (uPA) by tumours is an im portant pathway for neoplastic cell invasion and metastasis. Indeed in seve ral tumour types, elevated levels of uPA, its receptor (uPAR) or its inhibi tor plasminogen activator inhibitor-1 (PAI-1) is associated with a poorer p rognosis. Since endothelial cells also use this proteolytic system to remod el the extracellular matrix during angiogenesis and since angiogenesis, as assessed by microvessel density, is also a predictor of patient survival, t his study was designed to investigate the relationship between angiogenesis and the urokinase system in breast tumours. The aims were to assess whethe r the uPA, uPAR and/or PAI-1 correlates with angiogenic activity and could therefore be a useful objective clinical measure of tumour neovascularizati on; and to clarify whether the poor outcome associated with high levels of the urokinase system is due to its association with angiogenesis. The study also sought to examine the relationship between the uPA system and vessel remodelling using loss of a basement membrane epitope (LH39) normally assoc iated with established capillaries. The cytosolic levels of uPA, PAI-1 and uPAR were therefore measured by enzyme linked immunoabsorbent assay, togeth er with tumour vascularity, in 136 well-characterized invasive breast carci nomas. There were significant relationships between uPA and uPAR (Spearman r = 0.37, p < 0.0001), uPA and PAI-1 (Spearman r = 0.19, p = 0.03) and betw een uPAR and PAI-1 (Spearman r = 0.23 p = 0.01). A significant correlation was also observed between PAI-1 and vessel remodelling (Spearman r = 0.34, p = 0.04), patient age (p = 0.01), nodal status (p = 0.047) and tumour grad e (p = 0.04), but no association between tumour vascularity and PAI (p = 0. 96), uPA (p = 0.69) or uPAR (p = 0.81) was present. No significant associat ion was seen between any of the urokinase variables and expression of the a ngiogenic factor thymidine phosphorylase. Furthermore, no significant assoc iations were found between any of the studied parameters and overall surviv al in a univariate analysis of the cancer patients. A multivariate Cox prop ortional hazard model of overall survival showed that uPA (p = 0.15), but n ot uPAR (p = 0.52) or PAI-1 (p=0.61), gave no additional prognostic informa tion. These findings show that uPA may work via an independent pathway to a ngiogenesis and therefore combined blockade of uPA and angiogenesis may hav e additional therapeutic benefits. It also shows, as recently demonstrated in animal models, that PAI-I may be a key regulator of vascular remodelling in human cancer. Copyright (C) 2001 John Wiley & Sons, Ltd.