Predictors of pituitary tumour behaviour: an analysis from long-term follow-up in 2 tertiary centres

Objectives: To determine the clinical utility of assessment of tumour invasion, markers of proliferation, and the French clinicopathological classification in pituitary tumour prognostication. Methods: This is a retrospective evaluation of adult patients undergoing pituitary surgery at Oxford University and St Vincent's Hospitals, between 1989 and 2016, with at least 12 months of clinical data. Invasion was assessed radiologically, proliferative markers (Ki67, mitotic count, p53) by immunohistochemistry. Tumours were graded according to the clinicopathological classification. Intra- and interlaboratory variability of histopathology reporting was evaluated. Outcomes: (1) Tumour recurrence (radiological or reintervention ≥12 months postoperatively) and/or (2) "aggressive behaviour" (≥4 interventions and/or invasive tumour with recurrence/reintervention between 12 and 24 months postoperatively). Results: A total of 386 patients were included, age at surgery was 56 (interquartile range [IQR] 41-67) years, 54% were male, and median follow-up was 90 months (range 44-126). Tumours were predominantly clinically nonfunctioning (252, 65%), with overall 53% invasive, and 10% that demonstrated ≥2 proliferative marker positivity. Recurrence was predicted by invasiveness (hazards ratio [HR] 1.6 [1.10-2.37], P .02), elevated mitotic count (HR 2.17 [1.21-3.89], P .01), grade (2b vs 1a HR 2.32 [1.06-5.03], P .03), and absence of gross total resection (HR 3.70 [1.72-8.00], P .01). Clinically defined aggressiveness was associated with elevated Ki67, mitotic count, and invasiveness. Ki67 reporting methodologies showed moderate correlation across laboratories (Phi 0.620), whereas p53 reporting reproducibility was poor (Phi 0.146). Conclusions: Proliferative markers, including Ki67 and mitotic count, but not p53, are important in predicting the development of aggressive pituitary tumour behaviour.