Objectives: This study evaluates the diagnostic accuracy of SUV-based parameters derived from [(18) F]-2-fluoro-2-deoxy-D-glucose positron emission tomography (FDG-PET) in order to optimize non-invasive prediction of soft tissue tumour (STT) grade.
Methods: One hundred and twenty-nine lesions from 123 patients who underwent FDG-PET for primary staging (n = 79) or assessment of recurrence (n = 44) of STT were analyzed retrospectively. Histopathology was the reference standard for tumour grading. Absolute values and tumour-to-liver ratios of several standardized uptake value (SUV) parameters were correlated with tumour grading.
Results: At primary diagnosis SUVmax, SUVpeak, SUVmax/SUVliver and SUVpeak/SUVliver showed good correlation with tumour grade. SUVpeak (area under the receiver-operating-characteristic, AUC-ROC: 0.82) and SUVpeak/SUVliver (AUC-ROC: 0.82) separated best between low grade (WHO intermediate, grade 1 sarcoma, and low risk gastrointestinal stromal tumours, GISTs) and high grade (grade 2/3 sarcoma and intermediate/high risk GISTs) lesions: optimal threshold for SUVpeak/SUVliver was 2.4, which resulted in a sensitivity of 79 % and a specificity of 81 %. At disease recurrence, the AUC-ROC was <0.75 for each parameter.
Conclusions: A tumour SUVpeak of at least 2.4 fold mean liver uptake predicts high grade histopathology with good diagnostic accuracy at primary staging. At disease recurrence, FDG-PET does not reliably separate high and low grade lesions.
Key points: • Several SUV parameters accurately predict soft tissue sarcoma grade at primary diagnosis. • Tumour-to-liver ratios are of comparable value to absolute SUV parameters. • SUV peak /SUV liver >2.4 had 79 % sensitivity and 81 % specificity for high grade lesions. • At recurrence, FDG PET does not reliably indicate high grade sarcomas.