Adrenal lesions: variability in attenuation over time, between scanners, and between observers. (September 2013)

• Record Type: Journal Article
• Authors: Hammarstedt, Lilian; Thilander-Klang, Anne; Muth, Andreas; Wängberg, Bo; Odén, Anders; Hellström, Mikael
• Is Part Of: Acta radiologica. Volume 54:Number 7(2013)
• Page Start: 817
• Page End: 826
• Publication Date: 2013-09
• Subjects: Adrenal incidentaloma -- adrenal gland neoplasm -- computed tomography -- observer variation -- phantoms; Radiology, Medical -- Periodicals; Radiography, Medical -- Periodicals; Radiotherapy -- Periodicals; 616.0757
• DOI: 10.1177/0284185113482688
• Languages: English
• ISSNs: 0284-1851

Abstract

Background: Measurements of attenuation (in Hounsfield units [HU]) and contrast wash-out are widely used to characterize adrenal lesions as benign or indeterminate/malignant at computed tomography (CT). Clinical experience suggests that such measurements of adrenal lesions may vary over time and between observers, making evaluation difficult. Purpose: To investigate the change over time of adrenal lesion size, attenuation, and contrast wash-out at CT, to determine inter-observer variability, and to analyze other factors underlying the variability. Material and Methods: In a cohort of patients, with or without malignant disease, undergoing CT, adrenal lesions were prospectively analyzed. Lesions with growth >20% or >5 mm over 6 months were excluded. Non-enhanced attenuation and contrast medium wash-out over 2-year follow-up were analyzed. An inter-observer analysis with five observers and a phantom study of eight different CT scanners were performed to assess measurement variability. Results: Mean adrenal lesion non-enhanced attenuation values decreased by 0.5 HU/year during follow-up. Using 10 HU or 40% relative wash-out as threshold values for benign versus indeterminate lesions, 27 (20%) and 39 (29%) of 136 lesions, respectively, would be reclassified at some occasion during follow-up. In the observer analysis 37 of 40 lesions demonstrated agreement between all observers, using established threshold values. The phantom study showed an intra-scanner variability of 1–3 HU, but an inter-scanner variability of up to 8 HU for water. Conclusion: The clinically widespread use of specific attenuation threshold values for characterizing adrenal lesions must be used with great caution, considering that multiple factors, related to patient, equipment, scanning technique, and observer influence the outcome.