Inter- and intra-observer variability in detection and progression assessment with MRI of microadenoma in Cushing's disease patients followed up after bilateral adrenalectomy.

OBJECTIVE: To assess the inter- and intra-observer variability in detection and progression assessment with MRI of pituitary microadenomas in cases of Cushing's disease treated by adrenalectomy. DESIGN AND METHODS: Two radiologists independently reviewed initial and follow-up MRI studies of 32 patients with a history of Cushing's disease on two occasions each. Five classical qualitative signs and three quantitative measurements of microadenoma were considered. Intra and inter-observer reproducibility was evaluated with kappa statistics (qualitative signs) and intraclass correlation coefficients and Bland-Altman plots (quantitative measurements).
RESULTS: No sign or measurement could be identified as better than others and none displayed consistently high levels of inter- or even intra-observer reproducibility (kappa or intraclass correlation coefficient > 0.80). Both qualitative and quantitative criteria showed lower reproducibility when used for diagnosing progression than when used for diagnosing the presence of an adenoma. Reproducibility of qualitative signs appeared to be influenced by reader's experience.
CONCLUSIONS: Pituitary MRI study of patients with Cushing disease treated by adrenalectomy remains difficult. In practice, interpretation of MRI studies by a well-experienced reader should be preferred, unless a final decision can be reached by consensus between two and several readers.