Slamon 2001.

Study characteristics
Methods	Accrual time: June 1995 to March 1997 Multi‐centre, international Baseline comparability: balanced
Participants	469 female enrolled Age: trastuzumab + anthracycline arm: mean 54, range from 27 to 76; anthracycline control arm: mean 54, range from 25 to 75; trastuzumab + taxane arm: mean 51, range from 25 to 77; taxane control arm: mean 51, range from 26 to 73 Diagnosis: progressive metastatic breast cancer Inclusion criteria: women over‐expressing HER2 who had not previously received chemotherapy for metastatic disease. Only patients who had weak‐to‐moderate staining of the entire tumour cell membrane for HER2 (referred to as a score of 2+) or more than moderate staining (referred to as a score of 3+) in more than 10% of tumour cells on immunohistochemical analysis were eligible Exclusion criteria: patients were excluded if they had bilateral breast cancer, untreated brain metastases, osteoblastic bone metastases, pleural effusion or ascites as the only evidence of disease, a second type of primary cancer, or a Karnofsky score of less than 60, if they were pregnant or had received any type of investigational agent within 30 days before the study began HER2+: 100%
Interventions	4‐arm RCT Trastuzumab + anthracycline arm (randomised N = 143): doxorubicin (or epirubicin) plus cyclophosphamide (60 mg/m2 (75 mg/m2) and 600 mg/m2 every 21 days for 6 cycles) plus trastuzumab (first loading dose of 4 mg/kg, followed by weekly doses of 2 mg/kg until disease progression) Anthracycline control arm (randomised N = 138): doxorubicin (or epirubicin) plus cyclophosphamide (60 mg/m2 (75 mg/m2) and 600 mg/m2 every 21 days for 6 cycles) Trastuzumab + taxane arm (randomised N = 92): paclitaxel (175 mg/m2 every 21 days for 6 cycles) plus trastuzumab (first loading dose of 4 mg/kg, followed by weekly doses of 2 mg/kg until disease progression) Taxane control arm (randomised N = 96): paclitaxel (175 mg/m2 every 21 days for 6 cycles) Trastuzumab + anthracycline and anthracycline control arms: 100% patients naive to previous anthracycline chemotherapy Trastuzumab + taxane and taxane control arm: 97% patients previously treated with adjuvant anthracycline chemotherapy
Outcomes	Primary: time to disease progression (disease progression was defined as an increase of more than 25% in the dimensions of any measurable lesion) Secondary: OS, ORR, theduration of a response, the time to treatment failure (a compositeof disease progression, death, discontinuation of treatmentand the use of other types of antitumour therapy)
Notes	Study ID: HO648g Median follow‐up for efficacy: 30 months (30 min to 51 max) Upon documented disease progression patients (66%) were entered into the extension study H0659g, a non‐randomised, open‐label study in which they could receive either trastuzumab alone or in combination with chemotherapy of choice We found possible inconsistencies among different data reports (FDA Statistical Review ‐ Trastuzumab Product Approval Information ‐ Licensing Action 9/25/98, EMEA EPAR H‐C‐278 Scientific discussion for the approval of Herceptin 2004). For this systematic review we considered the NEJM publication as our primary data source OS: HR estimated using the ratio of the medians. HR variance estimated using the relationship between Chi2 and the log of the HR. Time to progression: HR estimated using the ratio of the medians. HR variance estimated using the relationship between Chi2 and the log of the HR
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not reported Stratification on the basis of history of adjuvant anthracycline treatment
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding of participants and personnel (performance bias) Overall Survival	High risk	Open‐label
Blinding of outcome assessment (detection bias) (OS)	Low risk	Open‐label (original double‐blind design was abandoned due to ethical considerations), but our primary outcome (i.e. OS) is not likely to be influenced by lack of blinding
Blinding of outcome assessment (detection bias) (outcomes other than OS)	Unclear risk	Open‐label
Incomplete outcome data (attrition bias) All outcomes	Low risk	Enrolled 469 patients. 5 patients were never treated; 2 decline treatment; 1 died before treatment; 1 had disease progression at enrolment; 1 was enrolled inadvertently
Selective reporting (reporting bias)	Low risk	The protocol for the study is available (http://www.cancer.gov/clinicaltrials/search/view?cdrid=64329&version=HealthProfessional&protocolsearchid=6378103)