Background Retroperitoneal fibrosis (RPF) could be idiopathic or secondary to various factors including malignancy. Pathology remains the most reliable diagnostic approach. However, the pathological features have yet been well defined and the feasibility of biopsy procedure depends on several factors such as the proximity of the lesions to abdominal aorta. 18F FDG-PET/CT may help exclude malignancy by assessing FDG-uptake and mapping the whole body, especially when biopsy procedure is inaccessible.
Objectives To evaluate the utility of PET/CT in distinguishing benign from malignant RPF.
Methods Patients with benign or malignant RPF diagnosed between September 2011 and June 2016 were included. Morphologic features, FDG-uptake of retroperitoneal lesions and lymph nodes (LNs) were compared. FDG-uptake of retroperitoneal lesions was assessed visually with the referrence of liver and assessed semiquantitatively with SUVmax. LNs were regarded as specific LNs if frequencies of LNs with high FDG-uptake were observed differently between two forms of RPF. ROC analyses were performed to evaluate the diagnostic accuracy.
Results Seventy-one cases with benign RPF and 21 cases with malignant RPF secondary to lymphoma (15) or metastatic carcinoma (6) with primary sites of pancreas (2), colon (2), stomach (1) and kidney (1) were included. The craniocaudal length, axial width, and distances between anterior or posterior limits and aorta were less in benign RPF, however, significant differences were only observed when comparing with lymphoma (p values: all <0.001) but not with metastatic carcinoma (p value: 0.396, 0.181, 0.112 and 0.64). A greater frequency of retroperitoneal lesions with high FDG-uptake (100% vs 77.5%, p value: 0.017) and a higher mean SUVmax (12.2 vs 4.8, p value: <0.001) were observed in malignant RPF. The frequencies of LNs with high FDG-uptake were greater with significance in malignant RPF except for hilar/mediastinal and cervical LNs, hence the rest LNs were regarded as specific LNs. At ROC analyses, the AUCs of SUVmax and specific LNs were 0.893 and 0.947. The sensitivity and specificity were 85.7% and 81.4% when the SUVmax was 6.23. The AUC of logistic regression model combining SUVmax and specific LNs was 0.974 with sensitivity of 90.5% and specificity of 90.1%.
Conclusions PET/CT could help distinguish benign from malignant RPF, especially when taking into account the FDG-uptake of retroperitoneal process and LNs.
Disclosure of Interest None declared