Intra- and inter-observer agreement in the visual interpretation of interim 18F-FDG PET/CT in malignant lymphoma: influence of clinical information.

Abstract

Background Interim PET/CT is widely performed in lymphoma patients in clinical practice and clinical trials. Visual assessment using a 5-point scale is proposed for PET/CT interpretation, but intra- and inter-observer variation is not fully investigated. Purpose To investigate intra- and inter-observer variations in the reporting of interim positron emission tomography/computed tomography (PET/CT) in lymphoma patients, and the influence of clinical information on the interpretation. Material and Methods Three expert readers from different institutions interpreted interim PET/CT images of 42 consecutive patients with malignant lymphoma twice, with and without clinical information. The intra- and inter-observer agreements were calculated using the kappa statistic on a patient and a region basis. Results On a patient basis, intra-observer agreement, inter-observer agreement without information, and inter-observer agreement with information were within the ranges 0.48-0.62, 0.51-0.62, and 0.42-0.76, respectively. In the evaluation of lymph nodes, intra-observer agreement, inter-observer agreement without information, and inter-observer agreement with information were within the ranges 0.78-0.92, 0.80-0.82, and 0.77-0.83, respectively. Observer agreements were in almost perfect to substantial agreement categories for most lymphatic organs, but were generally low for the other organs. Conclusion The intra- and inter-observer agreements in evaluating interim PET/CT were relatively low for extranodal lesions, but they were substantial to almost perfect when interpreting nodal regions in malignant lymphoma, irrespective of the provision of clinical information, although memory at the first interpretation might have affected the intra-observer results.