CORRESPONDENCE Re: Microarray Studies Challenge Theories of Metastasis

The recent evidence that tumors al-ready possess the potential to metas-tasize or have already done so when the primary tumor is removed has been interpreted as being incompatible with the current multistep carcinogenesis model (1). However, the model could still be true even if the molecular signa-tures for metastasis, independently pro-posed by van de Vijver et al. (2), Pomeroy et al. (3), and Singh et al. (4), are confirmed by larger studies. The multistep carcinogenesis model pro-poses that tumors start with benign growth and then develop the invasive and metastatic phenotype through the accumulation of somatic mutations. These mutations could occur at very early stages, long before the tumor is actually diagnosed with early detection methods such as mammography. Atypi-cal crypt foci, the very first sign of car-cinogenesis in the colon, are estimated to have already undergone thousands of mutations (5). Detecting tumors early does not necessarily mean that they are detected before they have completed most of the steps on the way to metas-tasis. It is now well established that a criti-cal step in tumor development is the acquisition of the ability to stimulate an-giogenesis in the tumor to provide oxy-gen and nutrients (6). Blood vessels also provide the major route for tumor cell dissemination. The process of angiogen-esis relies on many of the same mecha-nisms used for cell migration. The switch to the angiogenic phenotype occurs long before the tumor becomes clinically overt. Most of the multistep carcinogen-esis process, thus, could already have occurred when the tumor is detected, consistent with the recent microarray data. In addition to the need for a molecu-lar signature for treatment-related deci-sions, the emerging microarray data support the need for prevention strate-gies that target the angiogenic switch in specific high-risk populations (7)