Comparison of quantitative metabolic parameters of measurable tumors for the prediction of treatment response in patients with lymphoma using FDG PET

Abstract

1776

Objectives: The treatment response of lymphoma has usually been documented in measurable tumors using morphological cross-sectional imaging. However, it has been frequently reported that FDG PET yielded a more accurate assessment of therapeutic response than morphological imaging only. The aim of this study was to evaluate the adequate quantitative metabolic parameters for the prediction of therapeutic response.

Methods: FDG PET was performed in 35 patients with lymphoma (5 HD, 30 NHL) at pre-treatment and after 1 week of 1st cycle of chemotherapy. Therapeutic response was evaluated by CT scans as the recommendation of IWC after the completion of chemotherapy (4 to 8 cycle). On each FDG PET scan, maximum SUV (maxSUV) was measured at the measurable tumors. We also acquired volume of interest (VOI) by the automatic edge detection software at the cut-off value of 50% maximal voxel activity in the measurable tumors. According to the VOI, we measured the metabolic volume and mean SUV, and then, calculated volume-activity indexes (SUV*Vol) as mean SUV times metabolic volume. Then, we acquired the decrements (%) of each metabolic parameter on FDG PET after 1 week of 1st-cycle of chemotherapy. Diagnostic performance of each metabolic parameter decrement was evaluated by receiver-operating characteristics (ROC) curves for prediction of therapeutic response (CR or not).

Results: Of 35 patients, one patient with skin involvement was excluded because of the difficulty in locating measurable tumors. Initial stages of 34 lymphoma patients were stage I in 6, II in 11, III in 11, and IV in 6 by Ann Arbor system. Clinical responses after the completion of chemotherapy were CR/CRu in 30, PR in 1, PD in 3 by IWC. By the ROC curve analysis, area under the curve (AUC) and optimal criterion of each metabolic parameter decrement (%) according to the therapeutic response were as the table. There was no stastically significant difference between the metabolic parameters. Among the metabolic parameters, however, SUV*vol had the largest AUC (0.875) and optimal criterion of approximately 50% decrease for the prediction of treatment response.

Conclusions: By ROC curve analysis, metabolic parameters of measurable tumors on FDG PET could be applicable to predict treatment response in patients with lymphoma. Among the metabolic parameters, SUV*Vol showed the largest AUC by ROC curve analysis to predict treatment response at the optimal criterion of approximately 50% decrement.