In patients with locally advanced adenocarcinoma of the oesophagogastric junction (AEG), early metabolic response defined by 18-fluorodeoxyglucose-PET ([(18)F]FDG-PET) during neoadjuvant chemotherapy is predictive of histopathological response and survival. We aimed to assess the feasibility of a PET-response-guided treatment algorithm and its potential effect on prognosis.

Between May 27, 2002, and Aug 4, 2005, 119 patients with locally advanced adenocarcinoma of AEG type 1 (distal oesophageal adenocarcinoma) or type 2 (gastric cardia adenocarcinoma) were recruited into this prospective, single-centre study. All patients were assigned to 2 weeks of platinum and fluorouracil-based induction chemotherapy (evaluation period). Those with decreases in tumour glucose standard uptake values (SUVs), predefined as decreases of 35% or more at the end of the evaluation period and measured by PET, were defined as metabolic responders. Responders continued to receive neoadjuvant chemotherapy of folinic acid and fluorouracil plus cisplatin, or folinic acid and fluorouracil plus cisplatin and paclitaxel, or folinic acid and fluorouracil plus oxaliplatin for 12 weeks and then proceeded to surgery. Metabolic non-responders discontinued chemotherapy after the 2-week evaluation period and proceeded to surgery. The primary endpoint was median overall survival of metabolic responders and non-responders. Secondary endpoints were median event-free survival, postoperative complications and mortality, number of residual tumour-free (R0) resections, and histopathological responses. This study has been registered in the European Clinical Trials Database (EudraCT) as trial 2007-003356-11.

110 patients were evaluable for metabolic responses. 54 of these patients had metabolic responses (ie, decrease of 35% or more in tumour glucose SUV) after 2 weeks of induction chemotherapy, corresponding to a response of 49% (95% CI 39-59). 104 patients had tumour resection (50 in the responder group and 54 in the non-responder group). After a median follow-up of 2.3 years (IQR 1.7-3.0), median overall survival was not reached in metabolic responders, whereas median overall survival was 25.8 months (19.4-32.2) in non-responders (HR 2.13 [1.14-3.99, p=0.015). Median event-free survival was 29.7 months (95% CI 23.6-35.7) in metabolic responders and 14.1 months (7.5-20.6) in non-responders (hazard ratio [HR] 2.18 [1.32-3.62], p=0.002). Major histological remissions (<10% residual tumour) were noted in 29 of 50 metabolic responders (58% [95% CI 48-67]), but no histological response was noted in metabolic non-responders.

This study confirmed prospectively the usefulness of early metabolic response evaluation, and shows the feasibility of a PET-guided treatment algorithm. These findings might enable tailoring of multimodal treatment in accordance with individual tumour biology in future randomised trials.