18F-FDG PET for the diagnosis and grading of soft-tissue sarcoma: a meta-analysis.

UNLABELLED PET using (18)F-FDG is increasingly used for the diagnosis and grading of tumors. Several studies have been performed that evaluate the diagnostic and grading performance of (18)F-FDG PET for soft-tissue sarcoma, but each study has had a limited sample size. Therefore, we undertook a comprehensive meta-analysis of the evidence. METHODS Relevant studies were identified from MEDLINE and EMBASE. Diagnostic and grading performance were evaluated for qualitative visualization; standard uptake value (SUV, cutoffs of 2.0 and 3.0); and metabolic rate of glucose (MRG, cutoff of 6.0 micro mol/100 g/min). Quantitative data synthesis included independent weighting of sensitivity and specificity, construction of summary receiver operating characteristic curves, and pooled analyses. RESULTS The meta-analysis included 15 studies with 441 soft-tissue lesions (227 malignant, 214 benign). For diagnosis of malignant versus benign lesions, typical pairs of sensitivity and specificity estimates from the summary receiver operating characteristic curves were 92% and 73% for qualitative visualization; 87% and 79% for SUV 2.0; 70% and 87% for SUV 3.0; and 74% and 73% for MRG 6.0. Diagnostic performance was similar for primary and recurrent lesions. By qualitative interpretation, (18)F-FDG was positive in all intermediate/high-grade tumors (95% confidence interval [CI], 97.3%-100%), 74.4% (95% CI, 58.6%-85.9%) of low-grade tumors, and 39.3% (95% CI, 29.1%-50.3%) of benign lesions (including 11 of 12 inflammatory lesions). Using an SUV cutoff of 2.0, respective rates were 89.4% (95% CI, 79.4%-95.6%), 33.1% (95% CI, 15.6%-55.3%), and 19.1% (95% CI, 10.6%-30.5%). Limited data on comparisons with MRI and CT showed no differences against (18)F-FDG PET in diagnosing recurrent and metastatic disease. CONCLUSION (18)F-FDG PET has very good discriminating ability in the evaluation of both primary and recurrent soft-tissue lesions. (18)F-FDG PET may be helpful in tumor grading but offers inadequate discrimination between low-grade tumors and benign lesions.