PET scan for the detection of histological transformation of follicular lymphoma: A systematic review of diagnostic performance

Abstract

The strength of evidence supporting use of PET in the evaluation of suspected histological transformation (HT) of follicular lymphoma (FL) is unknown. We conducted a systematic review of studies reporting the diagnostic performance of ≥1 PET parameters for the detection of HT in patients with known FL. We searched PubMed for any study reporting ≥1 diagnostic performance metrics. Risk of bias was evaluated with the QUADAS2 tool. We included 7 studies encompassing 152 patients with a biopsy showing FL (or indolent non-Hodgkin lymphoma) and 111 with a biopsy confirming HT. Study designs and study populations differed substantially. PET methods were poorly reported and ¹⁸F-FDG dose was highly variable. Most studies were judged to be at high risk of bias in the patient and index test domains of QUADAS2. The diagnostic performance of 5 PET parameters were reported in at least one study but only SUVmax (n = 7) was reported in >2. Median SUVmax ranged from 9.2 to 10.9 in FL/iNHL and from 13.7 to 24.4 in HT. While SUVmax was consistently higher in the HT group, there was considerable overlap between the two groups and significant variability between studies. Area under the ROC curve for SUVmax to distinguish between FL/iNHL and HT ranged from 0.68 to 0.97. Sensitivity and specificity of the proposed cutoffs also varied widely (sensitivity ∼0.6 to 1, specificity ∼0.4 to 1). In conclusion, few studies — mostly small and potentially biased — have addressed this question. Although SUVmax is generally higher in HT than in FL, the diagnostic performance and optimal cutoffs remain unclear. Proposed SUVmax cutoffs should not be used to determine whether a patient has HT or to decide whether a biopsy should be obtained. For now, we encourage physicians to evaluate results of their own practice to devise a prudent workup of suspected.