Are metastatic tumours histologically similar to the primary? Result of a blinded experiment

There is an entrenched belief among pathologists, biologists and clinicians alike that metastatic tumours are histologically similar to the primary. Although this belief has never been scientifically tested, it forms the conceptual foundation upon which metastasis research has been based for decades. In order to scientifically test this dogma, we conducted a blinded experiment in which three highly experienced professor grade cancer pathologists were asked to predict the site of origin of 298 metastatic tumours arising at nine different sites. The accuracy of prediction of all three examiners was just under 50%. However, there was very little inter-observer agreement, and Kappa statistic generated the value of 0.27 which is much below the figure of 0.4 required to detect even a moderate level of concordance. In only 56 / 298 (18.8%) cases did the examiners agree with each others diagnosis. Thus the possibility cannot be excluded that the correct diagnoses made by the examiners had happened by play of chance. Our finding suggests that histological relationship between primary and metastatic tumours requires further investigation.