Methods	This trial utilised a straightforward randomisation comparing either pelvic irradiation or chemotherapy after initial surgical excision
Participants	A total of 475 women were randomised. Forty-one were deemed to be ineligible due to protocol violations such as myometrial invasion was less then 50% of the way through the myometrium, histology review revealed a sarcoma or rapid progression after enrolment. An additional 49 cases were excluded from analysis due to non-endometroid histology. The eligibility criteria for this study were International Federation of Gynecology and Obstetrics (FIGO- prior to 1998) stage IC-IIIC endometrial carcinoma with deeper than 50% myometrial invasion and absence of any prior chemotherapy, irradiation, or surgery for the treatment of any other cancer. Patients with stage II or III without deeper than 50% myometrial invasion were ineligible for this study. Patients were required to be under 75 years old, to have a WHO performance status of 0 to 3, and to have no residual visible tumour after an abdominal hysterectomy and bilateral salpingo-oophorectomy. Surgical staging consisted ideally of pelvic and/or para-aortic lymphadenectomy. 96% of women had a pelvic lymphadenectomy, 29% had a para-aortic lymphadenectomy
Interventions	The chemotherapy group received cyclophosphamide (333 mg/m2), doxorubicin (40 mg/m2), and cisplatin (50mg/m2) (CAP chemotherapy) every 4 weeks for 3 or more courses. Dose modifications of doxorubicin and cisplatin were as follows: a 25% reduction of both drugs was allowed for body weight less than 40 kg or age greater than 70 years old, and a 50% reduction was allowed in patients with G3 or G4 myelosuppression Pelvic irradiation was given in an open field using the anterio-posterior parallel opposing technique within 4 weeks of surgery. Forty-five to fifty Gray was given within 4 to 6 weeks, with 9 to 10 Gray of irradiation administered per week (5 working days per week). Subsequently, additional irradiations were performed in 11 cases (5.7%) with paraaortic lesions and in 6 patients (3.1%) who received brachytherapy
Outcomes	The primary endpoint was OS and secondary endpoints were PFS and the incidence of toxicity
Results (OS & PFS)	The PFS rate at 5 years was 83.5% in the PRT group and 81.8% in the CAP group. The hazard ratio was 1.07 (95% CI, 0.65 to 1.76:P = 0.73) The OS rate at 5 years was 85.3% in the PRT group and 86.7% in the CAP group. The hazard ratio was 0.72 (95% CI, 0.40 to 1.29; P = 0.27)
Results (site of recurrence)	Thirty-three recurrences (17%) occurred in the chemotherapy group compared to 30 recurrences (15%) in the radiotherapy group. The patterns of recurrence were similar in both treatment groups. Specifically, the incidence of intrapelvic recurrence sites, such as the pelvis or vagina, was 7.3% (14/192) in the chemotherapy group and 6.7% (13/193) in the radiotherapy group while the incidence of extrapelvic recurrence sites, such as the peritoneal cavity, liver, lung, para-aortic lymph nodes and others, was 16.1% (31/192) and 13.5% (26/193) respectively
Toxicity	G3 and G4 toxicities were experienced in 4.7% (9/192) of women given chemotherapy and 1.6% (3/193) of women allocated radiotherapy. Myelosuppression was the main problem for women given chemotherapy and bowel obstructions were the main complication in the radiotherapy group No treatment-related deaths occurred in either group.
Notes
Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The treatment allocation involved simple randomisation. Each participant was assigned by central telephone system The study groups were remarkably well balanced for patient characteristics including age, postmenopausal status, co-morbidity, type of hysterectomy, postoperative stage, tumour grade, myometrial invasion, lymphovascular space invasion, cervical involvement, parametrial invasion, peritoneal cytology, adnexal metastasis, pelvic lymph node metastasis, and para-aortic lymph node metastasis. Treatment was completed in 98.9% (184/186) and 97.3% (183/188) of the patients in the PRT and CAP groups, respectively
Allocation concealment (selection bias)	Unclear risk	Not specified
Blinding (performance bias and detection bias) All outcomes	High risk	It is not possible to conceal the treatment allocation from the participants
Incomplete outcome data (attrition bias) All outcomes	Low risk	The initial enrolment involved 475 women. Forty-one were ineligible due to myometrial invasion of less than 50%, histological diagnosis of sarcoma, or rapid progression of disease after enrolment. An additional 49 patients with non-endometrioid histology were excluded. As a result, 385 patients were eligible for this trial. Seven patients in the PRT group did not receive PRT and 4 patients in the CAP group did not receive CAP Treatment was completed in 98.9% (184/186) and 97.3% (183/188) of the patients in the radiotherapy and chemotherapy groups, respectively. Pelvic radiation was defined as completed when the total radiation dose reached 40 Gray. Chemotherapy was said to have been completed when the number of cycles reached three. The median total doses were 50 Gray of pelvic irradiation and 1309 mg/m2 cyclophosphamide, 120 mg/m2 doxorubicin, and 180 mg/m2 cisplatin. The median number of chemotherapy courses was 3, ranging from 1 to 7. The median duration of treatment was 5.1 weeks and 11.4 weeks in the radiotherapy and chemotherapy groups, respectively
Selective reporting (reporting bias)	High risk	The overall trial showed no significant difference with five-year PFS 83% versus 82%; OS 85% versus 87%. A subgroup analysis of 120 women in a high- to intermediate-risk group defined as (1) stage pTIc over 70 years old or with grade3 endometrioid adenocarcinoma or (2) stage pT2 or pT3a (positive cytology) was presented in the same publication. It suggested the chemotherapy group had a significantly higher PFS rate (83.8% versus 66.2%, log-rank test P = 0.024, hazard ratio 0.44) and higher OS rate (89.7% versus 73.6%, log-rank test P = 0.006, hazard ratio 0.24). Nevertheless this study fulfils the search criteria for the inclusion in this meta-analysis using the original unselected data
Other bias	Low risk	The broad inclusion criteria and subgroup analysis has been criticised but the authors have never deviated from their initial protocol and the groups are remarkably well matched However, the graphs depicting overall survival describe 122 women allocated radiotherapy but the graph related to PFS provides data on 123 women
Main problems with the trial as perceived by this analysis	High risk	Other trials have used higher doses of chemotherapy and used a 3 weekly cycle, not 4 weekly The other issue worth of note is the inconsistent data reporting. In the early ASCO presentations of this trial , 208 women were described as having low or intermediate risk disease and 67 women had high-risk disease. The full report published in 2008 has slightly different numbers. According to the manuscript, 190 women had low or intermediate risk disease, 120 had high intermediate risk group and 75 had high-risk disease Low intermediate risk was defined in the manuscript and as deeply invasive, well or moderately differentiated cancer. Of the 190 women reported in them anuscript, the progression-free survival was 94.5% and 87.5% (P = 0.11) and overall 5 year survival was 95% and 91% (P = .2) associated with radiotherapy and chemotherapy respectively. Earlier presentations described 89.6% and 87.2% The high intermediate risk group had deeply invasive poorly differentiated tumours and involved women over the age of 70 or they had stage 2 or 3a cancer. There were 120 women in the manuscript and there was significant progression-free survival advantage associated with chemotherapy HR = .44 (.20 to .97) High-risk disease was defined as stage 2 and 3a in the ASCO presentation of 67 women and they gained a substantial progression-free survival advantage (64.2% versus 84.5%; P = 0.06) and an overall survival benefit from adjuvant chemotherapy (80.3% versus 97.5% P = 0.02). This benefit was not confirmed in the final manuscript. In fact, the manuscript describes a progression-free survival rate in this group at 5 years of 78.6% and 64.4% (P = 0.17) and the overall survival rate at 5 years of 75.7% and 71.1% respectively. This is now a non significant trend in favour of radiotherapy. Presumably the data matured with time as more events were reached