The continuing pandemic of cancer deaths requires a reassessment of our basic assump- tions about the nature of cancer and how to control it. Directly bearing on this, there has been an explosion of new information about the tumour microenvironment, which is a complex system of many cell types, includ- ing endothelial cells and their precursors, pericytes, smooth-muscle cells, fibroblasts of various phenotypes, myofibroblasts, neu- trophils and other granulocytes (eosinophils and basophils), mast cells, T, B and natural killer lymphocytes, and antigen-presenting cells such as macrophages and dendritic cells. All these cells can participate in tumour pro- gression. If the process of carcinogenesis and its end result, invasive and metastatic cancer, are viewed as the maladaptive response of an entire tissue or organ to both genetic and epi- genetic stress 1–3 , then knowledge and control of the immediate microenvironment within a developing tumour become as important as the corresponding knowledge and control of the dysfunctional epithelial cells within that tumour. New data from studies on the tumour microenvironment suggest that we might need to revise the very definition of the term ‘carcinoma’ (currently defined in classical terms as a malignancy derived from epithe- lial cells), and that to control cancer in the future, we need to regard carcinogenesis and carcinomas as phenomena that occur in tis- sues, not in individual cancer cells. From this perspective, the microenvironment becomes an integral, essential part of the cancer. Therefore, it is necessary to consider the microenvironment of a cancer and its associ- ated abnormal epithelium as a functional whole. In addition to existing preventive efforts directed at dysfunctional epithelium, we propose that it is now essential to develop new chemopreventive measures targeted to the tumour microenvironment to control the process of carcinogenesis and prevent cancer. The implementation of this new approach, in addition to the accepted approach of targeting dysfunctional epithelial cells, becomes increasingly important as new data, summarized below, indicate that there might be molecular lesions in cells of the microen- vironment and in epithelial cells themselves. Although there are many excellent recent reviews on the tumour microenvironment, and it has recently been proposed as a target for therapy 4,5 , the application of chemopre- vention to control the tumour microenviron- ment during the early stages of carcinogenesis has so far not received concerted attention. This Perspective is not intended to be a comprehensive review; rather, it attempts to indicate new directions and to highlight new priorities for future research. Although the microenvironment is also involved in the genesis of leukaemias and sarcomas, this Perspective deals only with carcinomas (the most common forms of malignancy), which originate in epithelia and their associated stroma. New insights The concept that the microenvironment of a developing tumour is a crucial regulator of carcinogenesis was originally proposed by Paget in his famous ‘seed and soil’ hypothesis. Recent data indicate that car- cinogenesis and tumour angiogenesis result not only from the interaction of cancer cells with endothelial cells of various origin (vascular or lymphatic), but that surround- ing ‘normal’ stromal and inflammatory cells and tissue also have a crucial role in direct- ing the formation of the blood vessels that nourish a developing tumour. This newer version of an old hypothesis now enables us to consider that stromal cells and their associated matrix, as well as cells of the immune system, have important roles in tumour angiogenesis and carcinogenesis 6–9 . There is a discrete order of events in physiologically acute inflammation and repair 10 (FIG. 1a). However, these events become chaotically disorganized during chronic unresolved inflammation and carcinogenesis (FIG. 1b). This chaotic local microenvironment has led to the suggestion that tumours are ‘wounds that do not heal’ 11 . The constant disruption of homeostasis by proliferating epithelial cells produces a chronic inflammatory reaction, which is an abortive attempt to re-establish homeostasis through tissue remodelling 12 . However, the classic players in acute inflammation (granulocytes, macrophages, endothelial cells and fibroblasts) that ordinarily lead to the resolution of a wound through an orderly series of events, instead react para- doxically to the presence of dysfunctional epithelial cells by promoting their survival and replication 12 . This process includes inflammatory angiogenesis. Continuing angiogenesis driven by myeloid inflamma- tory cells is well recognized as an important component of chronic inflammatory disor- ders such as rheumatoid arthritis 13 ; recent data indicate that it can also have a key role in the progression of cancer 12,14,15 . Epithelial cells can become dysfunctional if their microenvironment is severely perturbed. This control of epithelia by their stroma is a default condition in normal tissues, as adult organs do not change composition, size or shape by uncontrolled remodelling. Therefore, rather than being a passive reaction to the cancer cell, the microenvironment might be a primary active factor in determining whether dys- functional epithelial cells will continue to grow and invade in a particular local milieu or, alternatively, merely become an indolent micro-hyperplasia or even be eliminated. Recent observations suggest that paracrine control in a developing tumour depends not so much on the intrinsic identity of its cellular components, but on the phenotypic expression of specific factors that either pro- mote or suppress carcinogenesis. For TUMOUR MICROENVIRONMENT — OPINION The tumour microenvironment as a target for chemoprevention Adriana Albini and Michael B. Sporn Abstract | New data indicate that primary dysfunction in the tumour microenvironment, in addition to epithelial dysfunction, can be crucial for carcinogenesis. These recent findings make a compelling case for targeting the microenvironment for cancer chemoprevention. We review new insights into the pathophysiology of the microenvironment and new approaches to control it with chemopreventive agents. The microenvironment of a cancer is an integral part of its anatomy and physiology, and functionally, one cannot totally dissociate this microenvironment from what have traditionally been called ‘cancer cells’. Finally, we make suggestions for more effective clinical implementation of this knowledge in preventive strategies. PERSPECTIVES NATURE REVIEWS | CANCER VOLUME 7 | FEBRUARY 2007 | 1