BackgroundPhysical activity (PA) before and after breast cancer diagnosis has been reported to be associated with lower mortality. However, whether changes in the activity after diagnosis impact prognosis is unclear and has not received much attention. This study aimed to examine pre- to postdiagnosis leisure-time PA and breast cancer prognosis.MethodsWe used data from the MARIE study, a prospective population-based patient cohort study of 3813 postmenopausal breast cancer patients, aged 50–74 at diagnosis, recruited from 2002 to 2005, re-interviewed in 2009, and followed up until June 2015. Prediagnosis PA was assessed at recruitment; postdiagnosis PA was assessed at re-interview in 2009. To examine pre- to postdiagnosis change in PA, women were categorized by pre- and postdiagnosis PA using a cut-off of 7.5 MET-h/week for meeting PA recommendations and combined into four groups: insufficiently active, increasingly active, decreasingly active, and sufficiently active. Cox regression models with delayed entry were used to assess associations between pre- to postdiagnosis patterns of PA and overall mortality (OM), breast cancer mortality (BCM), and recurrence-free survival (RFS). Additional analyses of pre- and postdiagnosis PA (no activity (reference), low activity, sufficient activity) with cancer outcomes, such as using a time-dependent model, were performed. In total, 2042 patients were included in the analyses.ResultsThere were 206 deaths (114 from breast cancer) after a median follow-up time of 6.0 years after the 2009 interview. Compared to insufficiently active women, increasingly active women were at lower risk of OM, BCM, and RFS (HR (95%CI) of 0.50 (0.31–0.82), 0.54 (0.30–1.00), 0.58 (0.40–0.84), respectively). In sufficiently active women, associations for OM (0.75 (0.48–1.15)), BCM (0.61 (0.33–1.13)), and RFS 0.80 (0.57–1.14)) were similar to increasingly active women but attenuated, and decreasingly active women were not at lower risk for OM (0.91 (0.61–1.36)), BCM (0.80 (0.45–1.42)), and RFS (1.04 (0.76–1.43)). In time-dependent analyses, sufficient activity vs. no activity was associated with better OM (0.73 (0.57–0.93)), BCM (0.64 (0.46–0.89)), and RFS (0.82 (0.68–0.99)). Low activity was not significantly associated with prognosis.ConclusionOur data support benefits for breast cancer prognosis in being physically active pre- and postdiagnosis particularly for women who were insufficiently active prediagnosis.