Differentiation of Benign Fluid Collections from Soft-Tissue Sarcomas on FDG-PET/CT

Objective: To assess the diagnostic performance of (18)F-FDG PET-CT in differentiating soft tissue sarcomas (STSs) from benign fluid collections (BFs). Materials and methods: Four readers independently reviewed 100 lesions on (18)F-FDG PET-CT and subjectively classified each lesion as an STS or BF and scored the spatial pattern of (18)F-FDG avidity (SP) of each on a 4-point ordered scale (thin, moderate, thick, solid). Results: Subjective assessment by readers allowed sensitive (91%-98%) differentiation of STSs from BFs, with lower specificity (59%-91%). The STSs had significantly higher SUV(max) (median 10.7, range: 2.0-33.7) than BFs (median 2.8, range: 1.1-12.3). Reader agreement in assessment of SP had average κ = 0.61 (range 0.46-0.70). Classification of thick or solid SP as STS yielded an inter-reader averaged sensitivity and specificity of 69% and 98%, respectively. The presence of thick or solid SP resulted in 14.1-fold increase in partial odds of STS. Each unit increase in SUV(max) resulted in 1.35-fold increase in partial odds of STS. The receiver operating characteristic (ROC) curves and 95% intervals for SUV(max) alone and SUV(max) + SP overlapped. The average subjective assessments for the four readers and estimated performance of using SP alone were both contained within the 95% intervals of the two ROC curves. Conclusions: (18)F-FDG PET-CT is a sensitive modality for differentiating STSs from BFs. SUV(max) and SP are significantly associated with STS. Classification schemes based upon SUV(max )alone or augmented with SP can be useful for distinguishing STS from BF.