Prediction of Response to Preoperative Chemotherapy in Adenocarcinomas of the Esophagogastric Junction by Metabolic Imaging

Abstract

PURPOSE: Preoperative chemotherapy in patients with gastroesophageal cancer is hampered by the lack of reliable predictors of tumor response. This study evaluates whether positron emission tomography (PET) using fluorine-18 fluorodeoxyglucose (FDG) may predict response early in the course of therapy. PATIENTS AND METHODS: Forty consecutive patients with locally advanced adenocarcinomas of the esophagogastric junction were studied by FDG-PET at baseline and 14 days after initiation of cisplatin-based polychemotherapy. Clinical response (reduction of tumor length and wall thickness by > 50%) was evaluated after 3 months of therapy using endoscopy and standard imaging techniques. Patients with potentially resectable tumors underwent surgery, and tumor regression was assessed histopathologically. RESULTS: The reduction of tumor FDG uptake (mean ± 1 SD) after 14 days of therapy was significantly different between responding (−54% ± 17%) and nonresponding tumors (−15% ± 21%). Optimal differentiation was achieved by a cutoff value of 35% reduction of initial FDG uptake. Applying this cutoff value as a criterion for a metabolic response predicted clinical response with a sensitivity and specificity of 93% (14 of 15 patients) and 95% (21 of 22), respectively. Histopathologically complete or subtotal tumor regression was achieved in 53% (eight of 15) of the patients with a metabolic response but only in 5% (one of 22) of the patients without a metabolic response. Patients without a metabolic response were also characterized by significantly shorter time to progression/recurrence (P = .01) and shorter overall survival (P = .04). CONCLUSION: PET imaging may differentiate responding and nonresponding tumors early in the course of therapy. By avoiding ineffective and potentially harmful treatment, this may markedly facilitate the use of preoperative therapy, especially in patients with potentially resectable tumors.