RESUMEN
PURPOSE: It has been suggested that FDG PET has predictive value for the prognosis of treated oesophageal carcinoma. However, the studies reported in the literature have shown discordant results. The aim of this study was to determine whether pretherapy quantitative metabolic parameters correlate with patient outcomes. METHODS: Included in the study were 67 patients with a histological diagnosis of oesophageal squamous cell carcinoma. Each patient underwent (18)F-FDG PET (4.5 MBq/kg) before chemoradiotherapy. Quantitative analysis was performed using the following parameters: age, weight loss, location, N stage, OMS performance status, MTVp and MTVp' (metabolic tumour volume determined by two different physicians), MTV40% (volume for a threshold of 40 % of SUVmax), MTVa (volume automatically determined with a contrast-based adaptive threshold method), SUVmax, SUVmean and TLG (total lesion glycolysis). RESULTS: MTVp and MTV40% were highly correlated (Pearson's index 0.92). SUVmeanp and SUVmean40% were also correlated (Pearson's index 0.86), as were TLGp and TLG40% (Pearson's index 0.98). Similarly, the parameters obtained with the adaptive threshold method (MTVa, SUVmeana and TLGa) were correlated with those obtained manually (MTVp, SUVmeanp and TLGp). The manual metabolic tumour volume determination (MTVp and MTVp') was reproducible. Multivariate analysis for disease-free survival (DFS) showed that a larger MTVp was associated with a shorter DFS (p = 0.004) and that a higher SUVmax was associated with a longer DFS (p = 0.02). Multivariate analysis for overall survival (OS) showed that a larger MTVp was associated with a shorter OS (p = 0.01) and that a tumour in the distal oesophagus was associated with a longer OS (p = 0.005). The associations among the other parameters were not statistically significant. CONCLUSION: Metabolic tumour volume is a major prognostic factor for DFS and OS in patients with oesophageal squamous cell carcinoma. Higher SUVmax values were paradoxically associated with longer survival. The location of the tumour also appeared to affect prognosis.