Comment www. thelancet. com/oncology Vol 18 March 2017 279 oesophagogastric cancer. 6 These data might be important for increasing the rate of curative resections and eventually survival outcomes of patients with localised oesophagogastric cancer in view of the disappointing R0 resection rates in ST03. The more proximal the anatomic tumour location, the worse the R0 resection rates tend to be. In the ST03 study, less than 70% of patients with lower oesophageal cancer had an R0 resection, with the circumferential resection margin being the crucial parameter for resection success. This percentage is probably unlikely to be improved by more radical surgery and we advocate for the addition of preoperative radiotherapy as investigated in the ongoing multinational TOPGEAR study. 7 The enrolment of 1063 patients with localised oesophagogastric cancer as done in the ST03 study is a massive undertaking. However, it might be reasonable to assess an established surrogate endpoint of survival, such as the R0 resection rate, after an enrolment of a representative subset of patients instead of reporting only on feasibility and toxicity as in the initial phase 2 component of ST03. 8 Adoptive study designs could prevent large negative phase 3 trials in the future. Additionally, treatment stratification according to tumour biology, 4 individual response, and risk features9 might increase the chances of new interventions being successful. Ultimately, despite the negative outcome of this trial, we hope that further developments in future studies will lead to more effective drugs for treatment of oesophageal and gastric cancer appearing on the horizon. 10