Unbiased profile development
Profiles for predicting tumor recurrence can be developed by comparing whole genome expression profiles of tumors that either metastasize or that do not recur. Those genes that are significantly different between the two tumor groups are probably the ones that can discriminate good and poor prognosis patients. The genes have been extracted in an unbiased way; there have been no human assumptions as to why certain genes end up in the profile. The next phase of a gene expression profile development is to validate whether the developed profile can be used in patient populations other that the patients the profile was developed in. It is the independent validation studies that determine the strength of a diagnostic profile.