Prognostic value of novel irregularity and heterogeneity measures of FDG uptake in osteosarcoma

Objectives To propose and evaluate novel measures of shape irregularity and heterogeneity of FDG uptake in osteosarcomas for prediction of outcome.

Methods FDG-PET had been performed prior to therapy in 17 osteosarcoma patients (17.7±9.7). The primary tumor was segmented using the ROVER 3D segmentation tool based on thresholding at the volume-reproducible intensity threshold after subtraction of local background. The novel measure of shape irregularity (IRREG) is defined as the deviation of the tumor's shape from sphere symmetry, quantified by the ratio of the third power of the tumor's surface to the second power of its volume. For the novel heterogeneity measure, SUVs were first binarized at a given threshold. Then a multiscale variance technique was used to compute heterogeneity of FDG uptake at the following spatial scales: 8, 16, 32mm. Heterogeneity was plotted as function of the binarization threshold for each scale. The areas under these curves (AUC) were computed. The ratio of AUC at 8mm to 16mm (HETER) was used for the univariate analysis presented here. Kaplan-Meier curves were obtained for IRREG, HETER, SUVmax and SUVmean with respect to both overall (OAS) and progression-free survival (PFS). Groups were separated by the best discrimination threshold according to ROC analysis and then compared by log-rank tests.

Results Median follow-up was 43 mo. Two patients died, 3 experienced tumor progression. Median PFS was 30.0 months. Both IRREG (p=0.0001) and HETER (p=0.05) provided prognostic power with respect to OAS. Higher pretherapeutic IRREG and HETER were associated with shorter OAS. Higher IRREG was associated with lower 3-years PFS rate (50% vs. 93%, p=0.02). Neither SUVmax nor SUVmean were predictive for OAS or PFS.

Conclusions The novel measures for heterogeneity and especially for irregularity seem promising for identification of high risk osteosarcomas