Since 1981, the cancer has occupied the first cause of Japanese death. Therefore, it is an urgent matter to tackle the cancer problem. Recently, it is understood that outcome of the cancer disease in a prognosis is related to the special combination of genes. However, it is very difficult to predict the outcome of the cancer disease in a prognosis, because a huge number of genes exist and we have to explore the special combination of genes among them. This study aims at finding the special genes combination, which affects on the outcome of the cancer disease in a prognosis. Self-Organizing Maps (SOM) is applied to accomplish our aim. Namely, SOM is used to cluster genes to predict the outcome of the cancer disease. Two analysis methods are also introduced into the SOM results, because it is insufficient for SOM only to specify the combination of genes, which affects on outcome of the cancer disease. Numerical experiments present some useful information on the outcome of the cancer disease in a prognosis.