Waddell 2013.

Methods	UK multicenter RCT (63 centers in UK), Phase III
Participants	Number of participants: 553 patients June 2008 to October 2011) Number randomised: Experimental group (epirubicin and oxaliplatin and capecitabine (EOC) + panitumumab): 278 control group (EOC): 275 Number evaluated: Experimental group: 278, age (median and range): 62 (26 ‐ 83) years Control group: 275, age (median and range): 63 (26 ‐ 83) years Diagnosis: patients with histologically‐verified, untreated, metastatic or locally‐advanced inoperable adenocarcinoma or undifferentiated carcinoma of the esophagus, gastro‐esophageal junction, or stomach. Inclusion: eligible patients should: be > 18 years; have measurable disease on CT or MRI; WHO performance status of 0 – 2; adequate cardiac, renal, liver, and bone marrow function
Interventions	Experimental group : epirubicin 50 mg/m² i.v. on day 1, every 3 weeks; oxaliplatin 100 mg/m² i.v. on day 1 every 3 weeks; oral capecitabine 1000 mg/m² per day on day 1 – 21 every 3 weeks; and panitumumab 9 mg/kg i.v. on day 1, every 3 weeks Control group: epirubicin 50 mg/m² i.v.on day 1 every 3 weeks; oxaliplatin 130 mg/m² i.v. on day 1 every 3 weeks; and oral capecitabine 1250 mg/m² per day on days 1 – 21 every 3 weeks Participants received a maximum of 8 cycles of treatment
Outcomes	Duration of follow‐up (median): 5.3 months (IQR: 2.6 ‐ 9.5) for experimental group, and 4.6 months (IQR: 1.8 ‐ 10.1) for control group OS, PFS, overall response, adverse events (only the incidence for grade ≥ 3 adverse events was reported, not for any adverse events)
Notes	Sponsor: Royal Marsden NHS Foundation Trust Study closed early due to lack of efficacy
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomisation was done independently at the Institute for Cancer Research Clinical Trials and Statistics Unit (ICR‐CTSU) by random permuted blocks (block sizes of six and eight) and stratified by centre region (locations were divided into 11 regions), extent of disease (locally advanced vs metastatic disease), and performance status (0 vs 1 vs 2)"
Allocation concealment (selection bias)	Low risk	"Patients were enrolled by trials office staff at the Royal Marsden Hospital, who then faxed confirmation of the allocated treatment group to local site staff"
Blinding of participants and personnel (performance bias) All outcomes	High risk	The open‐label design may introduce performance bias (e.g. through stress‐related mechanisms).
Blinding of outcome assessment (detection bias) OS, Serious adverse events	Low risk	These outcomes were unlikely to be affected by the blinding status of assessors
Blinding of outcome assessment (detection bias) PFS, Response, quality of life & adverse events	Unclear risk	No masking for outcome assessors, but quality of reported data was controlled by trust‐appointed monitoring staff or central monitoring system
Incomplete outcome data (attrition bias) All outcomes	Low risk	ITT analysis was used
Selective reporting (reporting bias)	Low risk	Prior protocol was available
Other bias	Unclear risk	Study closed early due to lack of efficacy