Prediction of tumor necrosis fraction using combined volumetric and metabolic indices derived from FDG-PET/CT in osteosarcoma patients receiving neoadjuvant chemotherapy

Abstract

10539 Background: Maximum standardized uptake values (mSUV) might not reliably reflect the chemotherapy response in osteosarcoma especially when treatment response within tumors is heterogenous. The purpose of this study was to compare the usefulnesses of various FDG PET/CT indices for predicting tumor response to neoadjuvant chemotherapy and to identify the most appropriate one in osteosarcoma. Methods: Thirteen patients with primary osteosarcoma (age 14±2.9 yrs, ranged 10–19 yrs) that had undergone FDG PET/CT scans before and after neoadjuvant chemotherapy were enrolled. The authors measured mSUV, metabolic tumor volumes (MTV), and total lesion glycolysis (TLG) in each PET/CT scan. MTVs were calculated by summing voxels with SUV greater than 1.5, 2, 2.5 and 3.0, and with % mSUV greater than 20%, 25%, 30% and 35%. Histopathologic necrosis fractions were compared with the above-mentioned PET/CT parameters and their pre- to post-treatment ratios (MTV ratio, rMTV; mSUV ratio, rSUV; TLG ratio, rTLG). Results: Histopathologic necrosis fractions ranged from 3% to 99% (62.2%±37.7%). rMTV and rTLG values were found to be correlated with histopathologic necrosis fractions (R2=0.45–0.65, p<0.05), whereas, mSUV and MR image volumes (MRV), both before and after treatment, rSUV values, and rMRV values were not. With regard to rMTV and rTLG values obtained using various MTV criteria, the highest correlation was observed for a rTLG value of 2.5 mg/ml (R2=0.65, p=0.001). Five patients were classified as responders and 8 as poor-responders to neoadjuvant chemotherapy defined as those with chemotherapy-induced necrosis fractions of ≥90% and <90%, respectively. The rSUVs of responders and poor-responders were not different, but rMTV and rTLG values showed a trend toward difference, although statistically not significant (p=0.110 and 0.106, respectively). The sensitivity and specificity of rTLG were 100% and 62.5%, respectively, at an SUV cut-off of 0.13. Conclusions: In our osteosarcoma patient population, TLG and MTV, which represent combined metabolic and volumetric indices, were found to predict tumor response better than pre- or post-treatment mSUV or rSUV. No significant financial relationships to disclose.