I‐SPY2.

Study characteristics
Methods	Accrual: May 2010 to July 2012 Multicentre, 30 sites Phase of trial: 2 Study design: open‐label adaptive/platform RCT Country or countries where the trial was conducted: USA Median follow‐up: not reported
Participants	Demographic and clinical characteristics presented below were not described separately for the hormone receptor‐negative cohort (i.e. 52% of the entire cohort). For the entire cohort Age: median 49, range 24–71 years Nodal status of breast cancer: 46% node positive, 54% node negative Adjuvant or neoadjuvant: neoadjuvant Notable exclusion criteria: none BRCA status collected: BRCA1/2 mutation – 17% in intervention arm, 7% in comparator arm
Interventions	Arm 1: paclitaxel 80 mg/m2 + veliparib 50 mg orally twice daily + carboplatin AUC6 every 3 weeks for 12 weeks followed by doxorubicin 60 mg/m2 + cyclophosphamide 600 mg/m2 every 2 or 3 weeks for 4 cycles Arm 2: paclitaxel 80 mg/m2 weekly for 12 weeks followed by doxorubicin 60 mg/m2 + cyclophosphamide 600 mg/m2 every 2 or 3 weeks for 4 cycles
Outcomes	Primary Probability of pCR, defined as the absence of residual cancer in the breast or lymph nodes at the time of surgery Secondary listed in trial publication and trial registry record (none reported in this paper) Predictive and prognostic indices to predict pCR and RCB 3‐ and 5‐year RFS 3‐ and 5‐year OS Adverse events (presurgery, postsurgery up to 1‐year follow‐up) MRI functional tumour volume RCB
Notes	Trial registration record: NCT01042379 Contacted for results for pCR in the triple negative subgroup. Only participants who received treatment were included in the analysis. Of those participants who did not receive treatment, it was unclear whether they were part of the TNBC cohort. Time‐to‐event outcomes collected but not yet reported. Funding considerations: funded by charitable donations and pharmaceutical companies including Johnson & Johnson, Genentech.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "… biomarker profiles are used for randomizing each participant to a treatment arm … for every participant that is randomized, there is a 20% chance the participant will be randomized to the control arm" as part of this adaptive trial design. Trial uses web‐based randomisation system.
Allocation concealment (selection bias)	Low risk	Randomisation was adaptive design with drug regimens being added or dropped depending on their efficacy. The registration and randomisation process was web‐based therefore it was unlikely that those involved in the trial were aware of intended treatment allocation.
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 involved regular laboratory tests, etc. at the end of each cycle. 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	A study‐trained pathologist evaluated pCR. The Study Lead Pathologist made the final assessment on any indeterminate or contested results. It was unclear whether the pathologists were blinded; however, pCR is generally viewed as an objective outcome and there was a minimal risk of treatment allocation affecting pCR assessment.
Selective reporting (reporting bias)	High risk	The primary outcome was reported; however, RFS and OS are yet to be reported. As pCR data were reported in 2016, other important long‐term efficacy outcomes would be expected to have been reported by 2022.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	CONSORT diagram was not provided for the cohort of interest in this review (i.e. TNBC); therefore, a judgement could not be made.
Other bias	Low risk	None identified.