Histological criteria for atypical pituitary adenomas - data from the German pituitary adenoma registry suggests modifications

Introduction: The term atypical pituitary adenoma (APA) was revised in the 2004 World Health Organization (WHO) classification of pituitary tumors. However, two of the four parameters required for the diagnosis of APAs were formulated rather vaguely (i.e., extensive nuclear staining for p53; elevated mitotic index). Based on a case-control study using a representative cohort of typical pituitary adenomas and APAs selected from the German Pituitary Tumor Registry, we aimed to obtain reliable cut-off values for both p53 and the mitotic index. In addition, we analyzed the impact of all four individual parameters (invasiveness, Ki67-index, p53, mitotic index) on the selectivity for differentiating both adenoma subtypes. Methods: Of the 308 patients included in the study, 98 were diagnosed as APAs (incidence 2.9 %) and 10 patients suffered from a pituitary carcinoma (incidence 0.2 %). As a control group, we selected 200 group matched patients with typical pituitary adenomas (TPAs). Cut-off values were attained using ROC analysis. Results: We determined significant threshold values for p53 (>= 2 %; AUC: 0.94) and the mitotic index (>= 2 mitosis within 10 high power fields; AUC: 0.89). The most reliable individual marker for differentiating TPAs and APAs was a Ki-67-labeling index >= 4 % (AUC: 0.98). Using logistic regression analysis (LRA) we were able to show that all four criteria (Ki-67 (p < 0.001); OR 5.2// p53 (p < 0.001); OR 3.1// mitotic index (p < 0.001); OR 2.1// invasiveness (p < 0.001); OR 8.2)) were significant for the group of APAs. Furthermore, we describe the presence of nucleoli as a new favorable parameter for TPAs (p = 0.008; OR: 0.4; CI95 %: 0.18; 0.77). Conclusions: Here we present a proposed rectification of the current WHO classification of pituitary tumors describing an additional marker for TPA and specific threshold values for p53 and the mitotic index. This will greatly help in the reliable diagnosis of APAs and facilitate further studies to ascertain the prognostic relevance of this categorization.