Mortality after breast cancer as a function of time since diagnosis by estrogen receptor status and age at diagnosis

Abstract

Our aim was to estimate how long-term mortality following breast cancer diagnosis depends on age at diagnosis, tumor estrogen receptor (ER) status, and the time already survived. We used the population-based Australian Breast Cancer Family Study which followed-up 1,196 women enrolled during 1992-1999 when aged <60 years at diagnosis with a first primary invasive breast cancer, over-sampled for younger ages at diagnosis, for whom tumor pathology features and ER status were measured. There were 375 deaths (median follow-up = 15.7; range = 0.8-21.4, years). We estimated the mortality hazard as a function of time since diagnosis using a flexible parametric survival analysis with ER status a time-dependent covariate. For women with ER-negative tumors compared with those with ER-positive tumors, 5-year mortality was initially higher (p < 0.001), similar if they survived to 5 years (p = 0.4), and lower if they survived to 10 years (p = 0.02). The estimated mortality hazard for ER-negative disease peaked at ~3 years post-diagnosis, thereafter declined with time, and at 7 years post-diagnosis became lower than that for ER-positive disease. This pattern was more pronounced for women diagnosed at younger ages. Mortality was also associated with lymph node count (hazard ratio (HR) per 10 nodes = 2.52 [95% CI:2.11-3.01]) and tumor grade (HR per grade = 1.62 [95% CI:1.34-1.96]). The risk of death following a breast cancer diagnosis differs substantially and qualitatively with diagnosis age, ER status and time survived. For women who survive >7 years, those with ER-negative disease will on average live longer, and more so if younger at diagnosis.

Keywords: breast cancer; cohort study; estrogen receptor; mortality; survival; time-dependent effects.

Figure 1.

Kaplan-Meier overall survival estimates from a population-based sample of women with breast cancer in the Australian Breast Cancer Family Study as a function of: estrogen receptor (ER) status of their tumors (ER-positive, blue; ER-negative, red; 95% confidence intervals, shaded area) (A); and age at diagnosis (<35, blue; 35–39, red; 40–49, green; 50–59, yellow) for ER-positive (B) and ER-negative (C) disease.

Figure 2.

Five-year mortality estimated from a population-based sample of women with breast cancer in the Australian Breast Cancer Family Study in terms of time since: they were diagnosed with breast cancer (A); if they survived to 5 years (B); and if they survived to 10 years (C). Estrogen receptor (ER)-positive, blue; ER-negative, red; 95% confidence intervals, shaded area.

Figure 3.

Predicted time-varying mortality hazard from a population-based sample of women with breast cancer in the Australian Breast Cancer Family Study at: age 35 years (A); age 45 years (B); and age 55 years (C) for estrogen receptor (ER)-positive (thick blue line) and ER-negative (thick red line) tumors. 95% confidence limits, thin blue and red lines.