Background The addition of adjuvant trastuzumab to chemotherapy has significantly improved outcomes in human epidermal growth factor receptor 2 (HER2) positive early, potentially curable breast cancer. Twelve months’ trastuzumab tested in the registration trials was adopted for standard adjuvant treatment in 2006. Subsequently similar outcomes were demonstrated using 9 weeks trastuzumab. Shorter durations were therefore tested for non-inferiority. Objectives To establish whether 6 months’ adjuvant trastuzumab is non-inferior to 12 months in HER2-positive early breast cancer using a primary endpoint of 4-year disease-free-survival (DFS). Design Phase III randomised, controlled, non-inferiority trial. Setting 152 NHS Hospitals. Participants 4088 patients with HER2-positive early breast cancer planned to receive both chemotherapy and trastuzumab. Intervention Randomisation (1:1) between six months’ or twelve months’ trastuzumab. Main outcomes Primary endpoint was DFS four years after diagnosis. Secondary endpoints were overall survival (OS), cost effectiveness, and cardiac function during trastuzumab. Assuming a 4-year DFS rate of 80% with 12 months, 4000 patients were required to demonstrate non-inferiority of 6-months (5% 1-sided significance, 85% power), defining the non-inferiority limit as no worse than 3% below the standard arm. Costs and quality-adjusted life years (QALYs) were estimated by within-trial analysis and a lifetime decision-analytic model. Results Between 4th October 2007 and 31st July 2015, 2045 patients were randomised to 12-months’ trastuzumab and 2043 to 6-months. Sixty-nine percent had ER-positive disease; 90% received anthracyclines (49% with taxanes; 41% without taxanes); 10% received taxanes without anthracyclines; 54% had trastuzumab sequentially after chemotherapy; 85% received adjuvant chemotherapy (58% were node negative). At 6.1 years median follow-up with 389 (10%) deaths, and 566 (14%) DFS events, 4-year DFS rates for the 4088 patients were 89.5% (95% CI, 88.1-90.8) in the 6-month group and 90.3% (95% CI 88.9- 91.5) in the 12-month group (Hazard Ratio 1.10; 90% CI 0.96–1.26, non-inferiority p=0.01), demonstrating non-inferiority of 6-months’ trastuzumab. Congruent results were found for OS (non-inferiority p=0.0003), and landmark analyses 6 months from starting trastuzumab (non-inferiority p=0.03 (DFS) and p=0.006 (OS)). 6-months’ trastuzumab resulted in fewer patients reporting adverse events of severe grade (365/1929 (19%) versus 460/1935 (24%) 12-month patients, p=0.0003) or stopping early because of cardiotoxicity (61/1977 (3%) versus 146/1941 (8%) 12-month patients, p<0.0001). Health economic analysis showed significantly lower lifetime costs and similar lifetime QALYs, and thus a high probability that 6 months is cost-effective compared to 12 months. Patient reported experiences on the trial highlighted fatigue, and aches and pains most frequently.

Limitations The type of chemotherapy and timing of trastuzumab changed through the recruitment phase of the study as standard practice altered. Conclusions PERSEPHONE demonstrated that in HER2-positive early breast cancer 6 months’ adjuvant trastuzumab was non-inferior to 12 months. There was significantly less cardiac toxicity and fewer severe adverse events with 6 months’ treatment. Future work On-going translational work investigates patient and tumour genetic determinants of toxicity, and trastuzumab efficacy. An individual patient data meta-analysis with PHARE and other trastuzumab duration trials is planned. Trial registration ISRCTN 52968807 Funding National Institute for Health Research, Health Technology Assessment Programme (HTA Project: 06/303/98).