AERO B03 2007.

Methods	Accrual: 12 December 2003 to 30 September 2004 Multicentre, phase II randomised controlled trial conducted in France Adjuvant therapy Median follow‐up: not reported
Participants	Age: median 53.9 years, range: 31–69 (arm 1) and 55.4 years, range: 32–72 (arm 2) 56–65% postmenopausal Node‐positive invasive breast adenocarcinoma Majority of women had T1‐2 and N0‐1 (arm 1: 91%; arm 2: 80%) Median number of pathologically involved nodes: 2 (range 1–20) ER‐positive: 68% (arm 1), 71% (arm 2); PR‐positive: 56% (arm 1), 45% (arm 2) HER2‐positive: 17% (arm 1), 6% (arm 2) Excluded: second or inflammatory breast cancer, previous or concomitant anticancer therapy including radiotherapy and hormone therapy
Interventions	3 arm study Arm 1 ('arm C' in the trial publication): docetaxel (100 mg/m²) every 2 weeks for 4 cycles followed by epirubicin (100 mg/m²), cyclophosphamide (600 mg/m²) every 2 weeks for 4 cycles. Arm 2 ('arm B' in the trial publication): epirubicin (100 mg/m²), cyclophosphamide (600 mg/m²) every 2 weeks for 4 cycles followed by docetaxel (100 mg/m²) every 2 weeks for 4 cycles. Arm 3 ('arm A' in the trial publication – not used in the review): docetaxel (75 mg/m²), epirubicin (75 mg/m²), cyclophosphamide (500 mg/m²) every 3 weeks for 6 cycles. For arms 1 and 2: pegfilgrastim (6 mg) recommended on day 2 after each chemotherapy cycle. Only arms 1 and 2 were used in the review
Outcomes	Primary outcome Incidence of grade 4 toxicity. Toxicity was assessed using the NCI CTC version 3 Secondary outcomes DFS OS Treatment adherence measures that included cumulative dose, absolute dose intensity (total actual dose received by patient, divided by number of weeks of treatment), and RDI (absolute dose intensity divided by planned‐dose intensity) Treatment‐related death
Notes	Trial not powered to detect differences between treatment arms. Clinical trial registration record not found. Trialists were contacted in July 2018 requesting information on efficacy outcome data; as of 16 August 2018, no reply received. Funding considerations: supported by European Association for Research in Oncology and by grants from Sanofi‐Aventis and Amgen.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Participants were "stratified according to center and number of involved lymph nodes (1‐3, 4‐9, > 9) and randomly assigned to one of three treatment arms." There were no imbalances in baseline characteristics.
Allocation concealment (selection bias)	Unclear risk	No details provided in trial publication.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	No information provided on blinding in the trial publication. Participants in both groups would have received the same drugs though in different order. Therefore, it was judged unlikely that this would lead to a material bias in participants' and physicians' behaviours even if unblinded.
Blinding of outcome assessment (detection bias) Overall survival	Low risk	OS data not yet reported. Unlikely that assessment of this outcome would be affected by unblinding.
Blinding of outcome assessment (detection bias) DFS	Low risk	Unblinding was unlikely to have an impact on outcome assessment for this review question where trial participants received both treatment regimens. Data for DFS have not been reported (results are not mature, as stated in 2007 publication).
Blinding of outcome assessment (detection bias) Toxicity and treatment adherence	Unclear risk	This study used formalised toxicity criteria (NCI CTC version 3) and measured a range of toxicity outcomes where in borderline cases some may be affected by unblinding (e.g. neuropathy). Dose‐reductions/delays were based on toxicity assessments. Overall, unblinding may have influenced the physicians' assessments for a subset of outcomes.
Incomplete outcome data (attrition bias) All outcomes	Low risk	31/34 (91%) participants in arm 1 and 25/31 (81%) participants in arm 2 completed the study. The reasons for not completing the study were similar across groups.
Selective reporting (reporting bias)	High risk	Did not identify clinical trial registry record; trial commenced in 2003 and was completed in 2004. Important efficacy outcome data – OS and DFS – were collected but not reported in trial publication published in 2007 (i.e. immature at time of publication). No data published in last 10 years. Trial authors were contacted in July 2018 and no reply received.
Other bias	Low risk	None identified