GOG 184.

Study characteristics
Methods	Multicentre parallel RCT conducted in the USA. Recruitment from 2000 to 2004.
Participants	659 women were enrolled. Women with FIGO stage III/IV endometrial cancer of any histology, with disease limited to the pelvis and abdomen were eligible.Women were not eligible if they had recurrent disease, previous radiotherapy, previous malignancy within five years, serious comorbidity, and survival expected to be < three months. All women received CRS and volume‐directed radiation to pelvic lymph nodes, with or without para‐aortic lymph nodes. CRS included hysterectomy and BSO. Radiotherapy was initiated within eight weeks of CRS. After the results of GOG 122 were known, women with stage IV were no longer eligible as GOG 122 showed a significant benefit with adjuvant chemotherapy after surgery in this group, compared with adjuvant radiotherapy.
Interventions	Chemotherapy was initiated within eight weeks of radiotherapy. Arm 1 (282 women) = Day 1:cisplatin (50 mg/m2)/doxorubicin (45 mg/m2)/Day 2:paclitaxel (160 mg/m2), every 21 days for six cycles. G‐CSF given to all women. Arm 2 (270 women) = Day 1:cisplatin (50 mg/m2)/doxorubicin (45 mg/m2) every 21 days for six cycles. G‐CSF initially optional, then in 2002 it was included in the control regimen for all women.
Outcomes	RFS (3 year), OS, loco‐regional recurrence, distant recurrence, adverse events
Notes	OS data were not mature at the time of the 2009 report. We were unsuccessful in obtaining these mature data from the investigators. Baseline characteristics overall, were: 69% endometrioid tumours; 12% stage IV; 41% grade 3; 90% absent or microscopic residual disease; 99% performance status 0‐1; median age 58 years. Median follow‐up was 47 and 46 months for the trial arms, respectively.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Central random assignment by the GOG Statistical and Data Center after CRS and radiotherapy. Randomisation was in balanced blocks.
Allocation concealment (selection bias)	Low risk	Randomisation was concealed from institutions and patients until centrally allocated.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	An unblinded trial; however, performance bias is unlikely to have an impact on survival data.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Central review of data and pathology materials was performed and a central review of eligibility data was conducted by the GOG Gynecologic Oncology Committee without knowledge of outcome.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Missing data was < 10%.
Selective reporting (reporting bias)	Low risk	All expected outcomes were reported.
Other bias	Low risk	A well‐designed and conducted trial. ITT analyses.