Despite significant progress in our understanding and treatment of metastatic cancer, nearly all metastatic cancers are incurable. In this Review, we use breast cancer as a model to highlight the limitations and inconsistencies of our existing treatment paradigms for metastatic disease. In turn, we offer a new theory of metastasis, termed "self-seeding. " The self-seeding paradigm, well validated in mathematical, experimental and animal models, challenges the notion that cancers cells that leave a primary tumor cell, unidirectionally seed metastases in regional lymph nodes and/or distant sites. In contrast, there is mounting evidence that circulating tumor cells can move multi-directionally, seeding not only distant sites but also their tumors of origin. Here, we show that the self-seeding model may answer many of the quandaries intrinsic to understanding how cancer spreads and ultimately kills. Indeed, redirecting our research and treatment efforts within the self-seeding model may offer new possibilities for eradicating metastatic cancer.
