Once the diagnosis of cancer is established, the most urgent question is whether the disease is localized or has spread to lymph nodes and distant organs. The most fearsome aspect of cancer is metastasis, and this fear is well justified. In nearly 50% of patients, surgical excision of primary neoplasms is not curative because metastasis has occurred by that time (Sugarbaker et al., 1977; Sugarbaker, 1979; Fidler and Balch, 1987). Often the metastases are too small to be detected (< 5 mm in diameter), and the primary neoplasm is surgically resected with curative intent. Unfortunately, the clinical reality is quite different. While the metastases grow only gradually in most patients, in a few, especially those with advanced local disease, surgical resection of the primary neoplasm can be followed by a rapid growth of visceral metastases. The latter phenomenon has troubled surgeons since the early 1900s and ha8 prompted many investigations of the factors regulating this process (for review see Sugarbaker et al., 1977). Over the last 80 years, many different hypotheses have emerged. These include the mechanical release of a large number of tumor cells during the surgical procedure, the sudden availability of “nutrients” for growth of metastases, the immunosuppressive effects of anesthesia and surgery that facilitate the escape of tumor cells from surveillance mechanisms, increased adhesive properties of platelets and blood coaguability, which aid the survival of circulating tumor emboli, and the production of a mitotic inhibitor by the local tumor (Sugarbaker et al., 1977; Gorelik, 1983; Prehn, 1993). The existence of so many divergent hypotheses suggests that the precise mechanism by which a primary neoplasm can control the growth of metastases is controversial. With a better understanding of the pathogenesis of cancer metastasis in general and the role that angiogenesis plays in this process in particular, the mystery is finally being solved.

CYReilly et al.(1994 [this issue of Ce//)) report the isolation and characterization of an antiangiogenesis factor produced by the transplantable murine Lewis lung carcinoma (3LL) growing in syngeneic mice. The angiogenic factor is a 38 kDa fragment of plasminogen that has been named angiostatin. Angiostatin has a long half-life in the circulation and can act in a paracrine or endocrine manner to inhibit the proliferation of endothelial cells. Turnover of the tumor vasculature is determined by the balance between stimulating and inhibiting molecules (Folkman, 1988, 1989; Folkman and Klagsbrun, 1987). Thus, when the resection of a local tumor suddenly depletes or eliminates circulating angiostatin, endothelial cells surrounding small metastases can proliferate and the tumor mass can expand. O’Reilly et al.(1994) also report that the systemic administration of angiostatin, but not intact plasminogen,