Visual scoring system for the assessment of treatment response in patients with oesophageal and gastro-oesophageal junction carcinoma

Aim: To evaluate the prognostic value of a 5 point FDG-PET/CT visual scoring system in patients with newly diagnosed oesophageal or gastrooesophageal junction (GOJ) cancer that have received neoadjuvant or definitive radiotherapy and/or chemotherapy. Methods: Ninety seven patients with endoscopically confirmed oesophageal or gastro-oesophageal junction carcinoma treated with definitive or neoadjuvant therapy received pre-treatment PET-CT scans, and of these, sixty seven received post-treatment PET/CT scans. Primary tumours were assessed using a 5 point visual scoring system (visual score of the primary tumour: PVS) for both the pre and post-treatment scans. The score is derived from comparisons between tumour uptake and background references adapted from the Deauville score used to assessment treatment response in lymphoma. Cox regression was used to analyse the relationship between pretreatment PVS (prePVS), post-treatment PVS (postPVS), change in PVS (deltaPVS) and survival (overall survival: OS, progression free survival: PFS). Log-rank tests were used to separate PVS into two prognostic categories with the greatest survival separation. Tumours were independently scored by 4 experienced observers blinded to all other information. Agreement was assessed using Cohen's kappa for individual scores and for two categories of scores (postPVS 1-3 vs 4-5 and delta PVS (-4)-(-2) vs (-1)-1). Results: On multivariate analysis postPVS and deltaPVS were significant predictors of survival. Patients with postPVS of 1 to 3 had significantly greater median OS (ranging from 53.3-56 months between all observers) compared to patients with post PVS 4-5 (15.2-23.6 months; all observers). Patients with primary tumours experiencing a reduction of 2 or more points had much greater median OS (56months - not reached in study; all observers) compared to patients whose tumours only experienced a point reduction of 1, had no change, or increased in points by 1 (15.1-21.8 months; all observers). Agreement for high uptake at site of primary tumour in the post-treatment scan (postPVS = 4-5) versus low uptake at site of primary tumour (postPVS = 1-3) was good with kappa ranging from 0.70 to 0.88 for paired combinations of all four readers. Conclusion: A visual assessment based on tumour uptake relative to reference background structures has the potential to provide valid and reproducible prognostic information following neo adjuvant or definitive radiotherapy +/- chemotherapy for oesophageal or GOJ carcinoma. These scores also have the potential to stratify treatment responders and nonre sponders, thus influencing management.