[Skip to Navigation]
[Skip to Navigation Landing]
Comment & Response
January 2019

Trastuzumab Therapy for 9 Weeks vs 1 Year for Human Epidermal Growth Factor Receptor 2-Positive Breast Cancer

Author Affiliations
  • 1Department of Biostatistics, Harvard T. H. Chan School of Public Health, Boston, Massachusetts
  • 2Department of Statistics and Actuarial Science, the University of Hong Kong, Hong Kong, China
JAMA Oncol. 2019;5(1):117-118. doi:10.1001/jamaoncol.2018.5730

To the Editor Joensuu and colleagues1 conducted an important study to assess the efficacy and safety of short-term (9 weeks) vs long-term (1 year) trastuzumab therapy for human epidermal growth factor receptor 2-positive breast cancer. The primary end point was disease-free survival (DFS) time. The noninferiority criterion was based on an assumed 5-year DFS rate of 85.0% for the 1-year group. An absolute 5-year DFS difference of less than 4% was considered clinically insignificant. In designing and analyzing the study, however, the authors converted 5-year DFS difference to the hazard ratio (HR) scale, resulting in a noninferiority margin of 1.3. Concerns about using HR for noninferiority studies have been discussed extensively.2-5 Because the authors chose to use HR, this trial was event driven. That is, 366 DFS events were needed to achieve desirable power for assessing noninferiority. However, the study ended with only 245 events. The 5-year DFS rates were 88.0% and 90.5% for the short-term and long-term treatment groups, respectively. The HR was 1.39 with a 90% CI of 1.12 to 1.72. It is well known that the precision of HR estimates depends on the number of events only, not on the patients’ exposure times. It is therefore no surprise that the CI is quite large.