Objective: We aimed to assess inter-observer agreement in bone involvement evaluation and define accuracy and reproducibility of MDCT images analysis in Multiple Myeloma (MM), by comparing two acquisition protocols at two different institutions.
Methods: A total of 100 MM patients underwent whole body low-dose computed tomography (WB-LDCT), with two protocols: Group I (50 patients), 80 kV and 200-230 mAs; Group II, 120 kV-40 mAs. Four readers (two experts) retrospectively reviewed 22 anatomical districts, reporting the following for each patient: 1) osteolytic lesions; 2) cortical bone integrity; 3) fractures; 4) risk of vertebral collapse; 5) hyperattenuating bone lesions; and 6) extraosseous extension. Inter-observer agreement (by all readers, expert and young observers and comparison of the two protocols) was then statistically analyzed.
Results: According to Cohen's criteria, inter-observer agreement among the four readers and between experts and residents was good for the detection of bone lesions and extra-medullary extension, and for the evaluation of risk of collapse and cortical integrity. There was good agreement when comparing the two protocols. A greater variability was found for the evaluation of hyperattenuating lesions and the presence of fractures.
Conclusions: WB-LDCT represents a reproducible and reliable technique that is helpful for defining bone disease in MM patients, with partial influence of readers' experience.
Key points: • MDCT represents a reproducible technique for defining bone disease in MM. • Overall inter-observer agreement is good, even when comparing two different protocols. • Influence of readers' experience on image analysis is partial.
Keywords: Agreement; Bone involvement; Inter-observer; Multidetector computed tomography; Multiple myeloma.