Technologies in molecular biology: Diagnostic applications

Abstract

For more than 100 years, cancer has been diagnosed by empirical, largely subjective means, not unlike clinical medicine itself. A consequence of this is that universal acceptance of a diagnosis, particularly any beyond benign versus malignant, has been an elusive goal. In one published study of rhabdomyosarcoma diagnosis involving more than 800 blinded cases and more than a dozen pathologists working in teams of two at eight institutions, with a 20% resampling, concordance between groups for subclassification (with clinical and therapeutic consequences) was no better than 60%. (The figure for individual reproducibility was marginally better, of the order of 70%.) It should be noted that these pathologists were acknowledged to be the world's experts in the diagnosis of this disease at the time. Thus, it is reasonable to conclude that cancer diagnosis is less than a perfectly objective science. © 2006 Springer Science+Business Media, Inc.