Susumu 2008.

Study characteristics
Methods	Multicentre RCT conducted in Japan. Recruitment from 1994 to 2000.
Participants	475 women were randomised but only 385 were analysed as 41 were found to be ineligible and 49 had non‐endometrioid histology. Inclusion criteria were age < 75 years, WHO performance status of 0‐3, FIGO stage IC ‐ IIIC with myometrial invasion > 50%. All women underwent CRS as primary treatment (TAH/BSO and, ideally, pelvic and paraaortic lymphadenectomy) and had no residual tumour. Patients were excluded if previously treated with surgery, chemotherapy or radiotherapy for another cancer.
Interventions	All women received CRS. Treatment was initiated within four weeks of surgery. Arm 1: adjuvant doxorubicin (40 mg/m2), cisplatin (50 mg/m2) and cyclophosphamide (333 mg/m2) every four weeks for three or more cycles. Arm 2: adjuvant external beam radiation therapy (45‐50 Gy on a five days per week schedule).
Outcomes	Five year OS, PFS and toxicity
Notes	Randomisation was not stratified by stage, however treatment groups had similar numbers of women with stage III disease (47 and 50 women in the radiotherapy and chemotherapy groups, respectively). Baseline characteristics of the groups were similar overall, including stage: 53% grade 3, 100% had myometrial invasion > 50%, 37% lymphovascular space invasion, 0% non‐endometrioid histology, mean age 59 years. We did not obtain separate baseline data for stage III women. Median follow‐up was 60 months. Only 97 women (25%) had stage III endometrial cancer. The investigators define a high risk group of 75 women with IIIB/C, or stage IIIA with other risk factors. From the trial publication, it is not clear how many of these women were allocated to each trial group. However, we have used the survival data in our analyses in the meantime and have requested complete stage III data from the trial investigators.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Simple randomisation was performed. "Each participant was assigned by a central telephone system".
Allocation concealment (selection bias)	Low risk	Assigned by a central telephone system.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	An unblinded trial (blinding not possible due to the nature of the interventions).
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Method of detection of progression not described; however, the primary outcome (OS) is unlikely to be affected.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	49 participants (10%) were excluded after randomisation due to non‐endometrioid histology (27 and 22 in RT and CT groups respectively) and 41 were deemed ineligible (18 and 23 respectively). This might be explained if randomisation was performed before surgical staging.
Selective reporting (reporting bias)	Low risk	All expected outcomes were reported. We obtained unpublished subgroup data from the investigators on the 21/3/2014.
Other bias	Low risk	None noted. ITT analyses.

RCT = randomised controlled trial; CRS = cytoreductive surgery; BSO = bilateral salpingo‐oophrectomy; FIGO = International Federation of Gynecology and Obstetrics; PFS = progression‐free survival; RFS = recurrence free survival; OS = overall survival; ITT = intention to treat.