[Bootstrapping algorithm approach reveals inherent regulatory pattern in 550 invasive breast cancer cases: CK5/6/CK14 and CK8/18/CK19 builds an antagonistic set].

All the preliminary observations on a lot of marker sets defining different stages in the tumor development are building a framework of work hypothesis which can be verified in characterising large pools of histological uniform rated paraffin probes. We developed a bootstrapping algorithm based on correlation measures to uncover regulatory patterns of immunohistochemical characterized tissue arrays with 550 invasive breast cancer cases. The algorithm is implemented in 'S' a computer language used to model mathematical solutions. Focussing on the Cytokeratins versus a set of prominent markers in breast cancer differentiation it will be obvious that markers which are known to appear in early (progenitor) forms conform to CK5/6 and CK14 while others associated with late stages conform to CK8/18 and CK19. Markers examined are among others EGFR, EMA, erb-B2, Vimentin, p53, ER and PR. The developed approach is an elegant and complete procedure to reveal the real regulatory patterns which are enclosed in a certain experimental design. The statistical significance of the results calculated by our algorithm is generally high and in the presented experimental design smaller than 0.6 * 10E-6.