Poor agreement in recognition of abnormal mitoses: requirement for standardized and robust definitions

Aims: The finding of abnormal mitoses is a helpful feature in differentiati ng between benign and malignant neoplasia and has prognostic significance f or some tumours. As the use of a histopathological variable is limited by t he reproducibility of its recognition, we tested the interobserver agreemen t in the classification of abnormal mitoses among histopathologists. Methods and results: Ten practising histopathologists were shown 30 potenti al mitotic figures and were asked to classify these as 'normal mitoses','ab normal mitoses' or 'not mitoses' according to the criteria each pathologist used in their routine practice. The results were analysed using kappa stat istics. Overall agreement was only fair with a combined kappa of 0.31 and t here was unanimous categorization of only four of 30 test items, none of wh ich as called abnormal, The poorest result was obtained for the category 'a bnormal mitosis' with only slight agreement (kappa 0.19). Agreement for the other categories varied from moderate (kappa = 0.45) for 'not a mitosis' t o fair (kappa = 0.26) for 'normal mitosis', Comparison of the results for o bserver pairs showed that for 12 out of the 45 possible pairings, there was no more agreement than might be expected by chance alone, Conclusion: Agreement is poor among practising histopathologists in the rec ognition of abnormal mitoses. A standardized and robust definition is neede d if diagnostic and prognostic significance is accorded to the finding of a n abnormal mitosis in the context of neoplasia.