Abstract
110
Objectives Diagnosis and prognosis is usually carried out manually and visually based on maximum SUV. The objective was to investigate the prognostic value of tumor volume (TV) and total glycolytic volume (TGV) measured by various methodologies in esophagus cancer patients.
Methods 26 esophagus cancer patients treated with radiochemotherapy between 2003 and 2007 were retrospectively considered and classified as complete, partial or non responders according to the EORTC recommendations. Various parameters were measured on the 18F-FDG pre treatment scan, using a fixed threshold at 42% of the maximum, an adaptive thresholding methodology with two different users and the Fuzzy Locally Adaptive Bayesian (FLAB): the maximum SUV value within the tumor (SUVmax), the tumor volume (TV) and the associated mean SUV (SUVmean), and TGV (TGV = TV multiplied by SUVmean). The prognostic value of each parameter was investigated using Keiplan-Meier survival curves and Kruskal-Wallis tests for correlation with response to therapy classification.
Results SUV measurements correlated poorly with therapy response or survival. On the other hand, TV allowed significant group differentiation for both survival and response, but only if measured using FLAB. TGV had better prognostic value and allowed significant differentiation for both response and survival and for all delineation methods, with however a better differentiation for FLAB than the threshold-based approaches.
Conclusions Our results suggest that the tumor volume information has significantly greater predictive value than SUV measurements only for prognosis of patients with esophagus cancer, as long as it is accurately measured. TGV is even better and is in addition less dependent on the method used to delineate the volume, although FLAB offered the best results compared to threshold-based methodologies. Future studies will investigate other cancer types as well as additional esophagus cancer patients in order to confirm these preliminary results