CALGB 40603.

Study characteristics
Methods	Accrual: May 2009 to August 2012 Multicentre Phase of trial: 2–3 Study design: open‐label RCT Country or countries where the trial was conducted: USA Median follow‐up: 7.9 years for EFS and OS data
Participants	Age: mean and range not reported. 60% aged 40–59 years Nodal status of breast cancer: 52% node positive, 42% node negative, 7% missing Adjuvant or neoadjuvant: neoadjuvant Notable exclusion criteria: none
Interventions	4‐arm study that has been grouped here into 2 categories according to carboplatin use. Arm 1 (listed as arms 3 and 4 in the trial publication): paclitaxel 80 mg/m2 weekly + carboplatin AUC6, 3 weekly for 12 weeks followed by doxorubicin 60 mg/m2 + cyclophosphamide 600 mg/m2 every 2 weeks for 4 cycles ± bevacizumab 10 mg/kg every 2 weeks for 9 cycles Arm 2 (listed as arms 1 and 2 in the trial publication): paclitaxel 80 mg/m2 weekly for 12 weeks followed by doxorubicin 60 mg/m2 + cyclophosphamide 600 mg/m2 every 2 weeks for 4 cycles ± bevacizumab 10 mg/kg every 2 weeks for 9 cycles Where possible, we reported pair‐wise comparisons of arm 1 vs arm 3 (see CALGB 40603 – comparison 1 (without bevacizumab)), and arm 2 vs arm 4 (see CALGB 40603 – comparison 2 (with bevacizumab)).
Outcomes	Primary pCR breast – absence of residual invasive disease with or without ductal carcinoma in situ (ypT0/is) Secondary pCR breast/axilla – pCR breast and the absence of any tumour deposit > 0.2 mm in sampled axillary nodes (ypT0/isN0) Treatment delivery Treatment‐related toxicity, graded according to NCI CTCAE v4.0 RCB‐conversion from clinically node positive to pathologically node negative Conversion from breast‐conserving surgery ineligible to breast‐conserving surgery eligible RFS/EFS for up to 10 years, defined as time from random assignment to local, regional or distant recurrence, any second invasive cancer, or death from any cause Time to first failure for up to 10 years OS for up to 10 years, defined as time from random assignment to death from any cause Distant RFS, defined as time from random assignment to detection of metastatic disease or death attributed to disease progression
Notes	Trial registration record: NCT00861705 Not all randomised participants were included in intention‐to‐treat analysis – participants who withdrew consent before completing neoadjuvant chemotherapy were excluded from pCR analyses. Study authors appeared to have assessed proportional hazards assumption. Funding considerations: funded in part by NCI grants, Genentech, the Breast Cancer Research Foundation, and the American Recovery and Research Act. These sources were not involved in data analysis or preparation of manuscript.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Randomisation in 2 × 2 format, method not described.
Allocation concealment (selection bias)	Low risk	Study protocol stated participant registration and randomisation occurs through CALGB web‐based system. It is most likely that randomisation was centralised.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Participants and personnel were aware of treatment allocation. This may have been associated with some performance bias but it was not judged to be of serious concern given types of outcomes collected.
Blinding of outcome assessment (detection bias): DFS	Low risk	Lack of blinding unlikely to influence this outcome.
Blinding of outcome assessment (detection bias): OS	Low risk	Lack of blinding unlikely to influence this outcome.
Blinding of outcome assessment (detection bias): toxicity	Low risk	Toxicity outcomes graded using the CTCAE. Although the study was open‐label, grading symptoms using the CTCAE is standardised and, therefore, knowing treatment allocation may have had minimal effect on the grading of outcomes.
Blinding of outcome assessment (detection bias): neoadjuvant studies only: pCR	Low risk	Pathological response was determined locally, without central pathological review. It was not reported whether this pathologist was blinded to the treatment allocation, however pCR is viewed to be an objective outcome and unlikely to be influenced by knowledge of treatment allocation.
Selective reporting (reporting bias)	Low risk	All prespecified primary endpoints from the trial record were reported in the manuscript. Some secondary outcome measures, including radiographic response, clinical response, and incidence and severity of postoperative complications were not reported; however, these were not considered to be critical outcomes for this review topic. OS, a secondary outcome, was reported in a subsequent publication.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	CONSORT diagram showed that 11/454 participants who were randomised but did not receive treatment were not included in the efficacy analysis. Reasons for exclusions were not detailed.
Other bias	Low risk	None identified.