Time-dependent patterns of recurrence after early stage breast cancer: Preliminary observations and methodological issues
Journal Article
Overview
abstract
536 Background: For patients with lymph node-negative (n0) breast cancer, benefit from adjuvant therapy is established. However, absolute gains are smaller than in node-positive disease and thus more selective use is warranted, prompting development of risk profiling to identify those most (or least) likely to benefit. In addition to defining risk based on recurrence-free status to specific time landmarks, examination of the magnitude and changes in the risk (hazard) of failure over time may be informative. We investigate recurrence hazards in n0 breast cancer. Methods: In a cohort of 9,279 participants from 5 NSABP randomized trials (accrual 1982–1998) investigating cytotoxic chemotherapy (CTX) and tamoxifen (TAM) for n0 breast cancer, we examine recurrence-free survival curves and recurrence hazards over time. The latter presents analytic challenges, because empirical (i.e., nonparametric) estimates are too unstable to provide reliable inference on patterns and do not incorporate covariates, while traditional semi-parametric (i.e., Cox) and parametric survival models are too restrictive to permit hazard changes. We apply flexible extensions of these models to examine and compare time-varying hazard profiles. Results: In patients undergoing surgery only, distinctive recurrence risk patterns between estrogen receptor (ER)-negative (-) and ER- positive (+) patients emerge. We observe an earlier, larger hazard peak in ER- patients, but a more persistent hazard in ER+ patients and a crossover (hazard for ER+ > ER-) occurring around 48 months. Under adjuvant treatment (CTX in ER- and TAM or TAM+CTX in ER+), magnitudes of failure hazards are reduced but other inter-relationships are largely maintained. For example, while TAM decreases the early hazard in ER+ patients to a level comparable to CTX treated ER- disease, in later follow-up (>5 years) the ER+ hazard again exceeds that of ER- patients. Adding CTX to TAM results in hazard reductions in both early and later periods. Conclusions: Examination of recurrence hazards over time with our models reveals changes in risk that may have biologic and therapeutic strategy relevance. The proposed analytic approaches address shortcomings of standard methods and will be developed further. No significant financial relationships to disclose.
