ABSTRACT
BACKGROUND: The histological diagnosis of early lesions of mycosis fungoides (MF) is often difficult for dermatopathologists and prior studies have shown a low agreement rate among pathologists. An important reason for such difficulty may be the lack of specific histological criteria. METHODS: We tested a new method to interpret and report biopsies suspicious for MF. The method is based on a grading system reflecting the pathologist's degree of diagnostic certainty. A panel of four pathologists independently assessed a set of 50 biopsies suspicious for MF first without (Phase I) and subsequently with specific histological criteria (Phase II). A third Phase was carried out after a training session, using a new set of cases with corresponding immunophenotyping and gene rearrangement analysis. Weighted and unweighted kappa statistics were used to assess inter- and intra-pathologist variation. RESULTS: The consensus rate among pathologists improved from 48% in Phase I to 76% in Phase III. Overall precision weighted kappas increased from 0.630 in Phase I to 0.854 in Phase III, indicating excellent inter-pathologist agreement by Phase III. There was a significant association between the presence of an abnormal phenotype and/or T-cell clonality and a higher diagnostic score. CONCLUSIONS: The use of uniform criteria and training sessions can substantially improve the consensus rate among pathologists in the diagnosis of MF.