© The Author 2006. Published by Oxford University Press.
CORRESPONDENCE |
RESPONSE
Affiliations of authors: National Cancer Institute of Canada, Clinical Trials Group, Kingston, Ontario, Canada (DT, JLP); Mayo Clinic, Rochester, MN (JNI); The Division of Hematology-Oncology, Massachusetts General Hospital, Boston, MA (PEG)
Correspondence to: Paul E. Goss, MD, PhD, FRCPC, FRCP (UK), Massachusetts General Hospital Cancer Center, Cancer Center Administration, 55 Fruit Street, Cox Building, Suite 640, Boston, MA 02114 (e-mail: pgoss{at}partners.org).
We appreciate Dr Vakaet's compliments relating to our NCIC CTG MA17 clinical trial and would like to respond to the questions he raises regarding our updated report (1). Vakaet notes the differences in the number of patients at risk and hence the number of breast cancer events between our first (2) and updated reports and raises the question as to whether the completeness of data follow-up should have been taken into consideration before the Data Safety and Monitoring Committee (DSMC) made the decision to terminate our study at the first planned interim analysis. He is correct that the main difference between these two reports is the completeness of follow-up and other data in the final analysis. However, the number of events reviewed by the DSMC at the interim analysis was sufficient to conclude that there was a substantial treatment effect, the statistical significance of which exceeded the preset stopping boundaries. Our final analysis further confirmed the decision made by the DSMC, a decision that we continue to stand by, and we do not agree that the trial was stopped prematurely.
Vakaet also raises questions about the eligibility of some patients in our study and suggests that we report the distribution of the median time between initial diagnosis of breast cancer and random assignment to the trial as the 5th and 95th percentiles. As indicated in the first paragraph of the study population subsection of our JNCI article (1), there were 14 patients who were ineligible because of too short or too long tamoxifen exposure but were included in the analysis based on an intent-to-treat principle. This explains the out-of-range values in the time between initial diagnosis of breast cancer and random assignment and in a more relevant variable, duration of tamoxifen therapy. We have calculated percentiles in these two variables as suggested: the 1st and 99th percentiles for the time between initial diagnosis of breast cancer and random assignment were, respectively, 56.5 and 80.2 months; the 1st and 99th percentiles for the more relevant variable, duration on tamoxifen, were, respectively, 4.5 and 5.9 years. The 5th and 95th percentiles were, respectively, 59.8 and 73.4 months for time between initial diagnosis of breast cancer and random assignment and 4.7 to 5.4 years for duration of tamoxifen.
REFERENCES
(1) Goss PE, Ingle JN, Martino S, Robert NJ, Muss HB, Piccart MJ, et al. Randomized trial of letrozole following tamoxifen as extended adjuvant therapy in receptor-positive breast cancer: updated findings from NCIC CTG MA.17. J Natl Cancer Inst 2005;97:1262–71.
(2) Goss PE, Ingle JN, Martino S, Robert NJ, Muss HB, Piccart MJ, et al. A randomized trial of letrozole in postmenopausal women after 5 years of tamoxifen therapy for early-stage breast cancer. N Engl J Med 2003;349:1793–802.
Related Correspondence
CiteULike 
Connotea 
Del.icio.us What's this?
J Natl Cancer Inst 2006 98: 1162.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||