Interim 18FDG PET/CT (PET-2) helps predict outcome and tailor treatment in adults with Hodgkin Lymphoma (HL). In contrast, PET-2 data on pediatric HL are rare, with discordant results.
Visual analysis using Deauville Score (DS) was proposed to assess PET response. However a 5-point scale did not preclude inter-observer reproducibility issues. Alternative approaches were developed to improve the accuracy and reproducibility of PET-2, mainly based on PET semiquantitative parameters. We evaluated the clinical usefulness of PET-2, analyzing visually and semiquantitatively pediatric HL patients referred to a single center study.
18FDG PET/CT was performed in 27 pediatric HL patients at baseline (PET-0), after 2 cycles of chemotherapy (PET-2) and at the end of treatment. PET response assessment was carried visually according to the DS as well as semiquantitatively by use of absolute decrease in semiquantitative parameters from PET-0 to PET-2 (ΔsumSUVmax0-2, ΔsumSUVmean0-2, ΔsumMTV0-2, ΔsumTLG0-2) and the corresponding Response Indexes (RI%sumSUVmax0-2, RI%sumSUVmean0-2, RI%sumMTV0-2, RI%sumTLG0-2). Clinical response assessment was performed according the Cheson’s Revised Response Criteria considering patients as responders (R) or non-responders (NR). Mean follow up was 24 months. T-student test for unpaired groups was performed to compare PET semiquantitative parameters between R and NR. Chi-square and Fisher exact test were performed to evaluate the association among categorical variables. The prognostic capability of 18FDG PET/CT was calculated by ROC analysis and expressed as area under curve (AUC). A Cox regression model was built to evaluate potential prognostic factors among PET parameters.
5/27 (18%) patients were NR at the end of therapy based on clinical outcome and, among them, only one became R at 24 months follow-up; another one remained NR while the other two died. Visual assessment was: DS=1 in 14/27 (52%), DS=2 in 1/27 (3%), DS=3 in 4/27 (15%) and DS=4 in 8/27 (30%) patients. Differences between R and NR were statistically significant for ΔsumSUVmax0-2 (t=2.45, p=0.026) and almost statistically significant for ΔsumSUVmean0-2 (t=1.88, p=0.071). No significant difference was found for the other parameters. Any association among Deauville evaluation and outcome at the end of therapy was found (Fisher exact test p=0.136). The better AUCs resulted for ΔsumSUVmax0-2 (0.836; cut-off<12.5, sensitivity 80%, specificity 91%). Any PET parameters resulted as prognostic factors by Cox regression. These results may also depend on the low number of events (NR).
Semiquantitative analysis seems to be more accurate than visual analysis to interpret PET-2 and predict outcome in pediatric HL patients. In particular, ΔsumSUVmax0-2 appears to be the best PET parameters in predicting therapy response assessment in pediatric HL patients. However these encouraging results warrant further confirmation in larger series.