Josee-Lyne Ethier1, Geoffrey M Anderson2, Peter C Austin3, Mark Clemons4, Wendy Parulekar5, Lois Shepherd5, Lily S Trasiewicz5, Dongsheng Tu5, Eitan Amir6. 1. Cancer Centre of Southeastern Ontario, Kingston, Ontario, Canada. 2. Institute of Health Policy, Management and Evaluation, University of Toronto, Canada. 3. ICES, Toronto, Ontario, Canada; Institute of Health Policy, Management and Evaluation, University of Toronto, Canada. 4. The Ottawa Hospital Cancer Centre, University of Ottawa, Ottawa, Canada. 5. Canadian Cancer Trials Group, Kingston, Ontario, Canada. 6. Princess Margaret Cancer Centre, Toronto, Ontario, Canada. Electronic address: eitan.amir@uhn.ca.
Abstract
BACKGROUND: Many women diagnosed with early-stage hormone-sensitive breast cancer die of causes other than their breast cancer. These competing risks can create challenges in analysing and clearly communicating data on risk of breast cancer recurrence or death. Here, we quantify the impact of competing risks on estimates of disease recurrence and benefit from therapy. PATIENTS AND METHODS: Using data from the MA.27, MA.17 and MA.17R trials of adjuvant endocrine therapy in early breast cancer, we compared Kaplan-Meier (KM) and competing risk methods for disease-free survival (DFS) and distant recurrence-free survival (DRFS). Each trial was analysed separately. In KM analyses, participants were censored at the time of non-breast cancer death. Competing risk analyses comprised cumulative incidence functions in which non-breast cancer death was a competing risk. RESULTS: Non-breast cancer deaths were observed more often in older participants, in those with lower risk of breast cancer and after longer follow-up. Compared with conventional analyses, estimates of the proportion of participants with DFS or DRFS events were lower in competing risk analyses, with this difference increasing over the course of follow-up. The absolute treatment benefit was similar or modestly lower in competing risk analyses. CONCLUSION: Compared with KM methods, competing risk analyses result in lower estimates of DFS and DRFS events and similar or modestly lower absolute benefit from experimental endocrine therapy. Over a long time horizon, competing risk methods may be preferable to KM methods when estimating future risk of recurrence in early-stage hormone-sensitive breast cancer. CLINICAL TRIALS REGISTRATION: Clinicaltrials.gov; NCT00003140, NCT00754845, NCT00066573.
BACKGROUND: Many women diagnosed with early-stage hormone-sensitive breast cancer die of causes other than their breast cancer. These competing risks can create challenges in analysing and clearly communicating data on risk of breast cancer recurrence or death. Here, we quantify the impact of competing risks on estimates of disease recurrence and benefit from therapy. PATIENTS AND METHODS: Using data from the MA.27, MA.17 and MA.17R trials of adjuvant endocrine therapy in early breast cancer, we compared Kaplan-Meier (KM) and competing risk methods for disease-free survival (DFS) and distant recurrence-free survival (DRFS). Each trial was analysed separately. In KM analyses, participants were censored at the time of non-breast cancer death. Competing risk analyses comprised cumulative incidence functions in which non-breast cancer death was a competing risk. RESULTS:Non-breast cancer deaths were observed more often in older participants, in those with lower risk of breast cancer and after longer follow-up. Compared with conventional analyses, estimates of the proportion of participants with DFS or DRFS events were lower in competing risk analyses, with this difference increasing over the course of follow-up. The absolute treatment benefit was similar or modestly lower in competing risk analyses. CONCLUSION: Compared with KM methods, competing risk analyses result in lower estimates of DFS and DRFS events and similar or modestly lower absolute benefit from experimental endocrine therapy. Over a long time horizon, competing risk methods may be preferable to KM methods when estimating future risk of recurrence in early-stage hormone-sensitive breast cancer. CLINICAL TRIALS REGISTRATION: Clinicaltrials.gov; NCT00003140, NCT00754845, NCT00066573.