ADAPT‐TN.

Study characteristics
Methods	Accrual: May 2013 to January 2015 Multicentre, 48 sites Phase of trial: 2 Study design: RCT Country or countries where the trial was conducted: Germany Median follow‐up: not reported
Participants	Age: median 50, range 26–75 years Nodal status of breast cancer: node positive 26%, node negative 74% Adjuvant or neoadjuvant: neoadjuvant Notable exclusion criteria: none
Interventions	Arm 1: Nab‐paclitaxel 125 mg/m2 + carboplatin AUC2 days 1 and 8, every 3 weeks for 4 cycles Arm 2: Nab‐paclitaxel 125 mg/m2 + gemcitabine 1000 mg/m2 days 1 and 8, every 3 weeks for 4 cycles
Outcomes	Primary pCR, defined as absence of invasive tumour cells in breast and lymph nodes (ypT0/is ypN0) Secondary Toxicity, according to NCI CTCAE version 4.0 EFS, defined as time from registration to any invasive relapse, secondary malignancy or death from any cause OS – early response Ki67 decrease > 30% or < 500 invasive cells in the 3‐week serial biopsy
Notes	Trial registration record: NCT01815242 Not all randomised participants were included in the analysis; 5 excluded prior to receiving treatment (3 in intervention arm, 2 in comparator arm). Safety analysis involved people receiving ≥ 1 dose of trial medication. For DFS and OS, we estimated the hazard ratio using Tierney's method. Study did not report assessing the proportional hazards assumption. Funding considerations: funded by Celgene and Teva.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Randomisation in a 1:1 ratio, method to generate random sequence was not described.
Allocation concealment (selection bias)	Low risk	Quote: "Randomisation was performed centrally at West German Study group."
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): 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	pCR was assessed by a local pathologist only. It is 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 and secondary endpoints from the trial record were reported in the manuscript and subsequent abstract publications.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	CONSORT diagram showed that 5/336 randomised participants did not receive treatment and were excluded from analysis. Reasons for exclusions were detailed, including consent withdrawn, and violation of inclusion criteria. Additionally, pCR results for 12 participants were not reported.
Other bias	Low risk	None identified.