Proportion of false-positive lesions at interim and end-of-treatment FDG-PET in lymphoma as determined by histology: Systematic review and meta-analysis

Abstract

Purpose To systematically review and meta-analyze the proportion of false-positive lesions at interim and end-of-treatment 18F-fluoro-2-deoxy-D-glucose positron emission tomography (FDG-PET) in lymphoma using biopsy as reference standard. Materials and methods Medline was searched for original studies. Methodological quality of included studies was evaluated, and results were meta-analytically summarized using random effects (in case of interstudy heterogeneity [I2 ≤ 50%]) or fixed effects (in case of no interstudy heterogeneity [I2 > 50%]). Results Eleven studies, comprising 139 patients who underwent biopsy of an FDG-avid lesion during or after completion of antilymphoma treatment, were included. Overall methodological quality was moderate. The proportion of false-positive results among all biopsied FDG-avid lesions at PET performed during of after completion of treatment ranged between 7.7% and 90.5% (the vast majority was due to inflammatory changes), with a weighted summary proportion (random effects, I2 = 75.7%) of 55.7% (95% confidence interval [CI]: 32.6–76.6%). There were no available studies on interim FDG-PET in Hodgkin lymphoma. The pooled summary false-positive proportions were 83.0% (95% CI: 72.0%–90.2%) for interim FDG-PET in non-Hodgkin lymphoma (fixed effects, I2 = 27.7%), 23.1% (95% CI: 4.7%–64.5%) for end-of-treatment FDG-PET in Hodgkin lymphoma (random effects; I2 = 67.1%), and 31.5% (95% CI: 3.9%–83.9%) for end-of-treatment FDG-PET in non-Hodgkin lymphoma (random effects, I2 = 68.3%). Conclusion Both interim and end-of-treatment FDG-PET scans in patients with lymphoma suffer from a very high number of false-positive FDG-avid lesions. This finding, in combination with the previously reported high number of false-negative FGD-PET scans for residual disease detection, suggests that the role of interim and end-of-treatment FDG-PET should be reconsidered.