1018Long-term risk of distant metastasis and post-metastasis survival for breast cancer: health record linkage study

Background After early breast cancer (BC) treatment, women need information about long-term prognosis. In this population-based health record linkage study, we assessed the cumulative incidence of distant metastasis (DM) conditional on the DM-free interval; and BC-specific survival post-metastasis. Methods We included all women diagnosed with non-metastatic BC in the NSW Cancer Registry, 2001-2002. We used linked records from hospitals, dispensed medicines, radiotherapy services and death registrations and applied stringent criteria to determine time to first DM and BC death. Results 6338 women were included (BC: localised 3885, regional 2453). The 5-year cumulative incidence of DM was 7.4% (95% confidence interval 6.6-8.3) for localised BC; 22.8% (21.2-24.5) for regional BC. For women DM-free at 5 years, it was 5.7% (5.0-6.6); and 11.4% (10.0-13.0) to year 10, respectively. The annual hazard for BC death following localised BC remained lower than that for non-BC causes; for regional BC, it was similar to that for non-BC causes at 8 years. Following DM (N = 1492), BC-specific survival varied widely (median 25 months, interquartile range 6-127). The probability of surviving BC for ≥5 years was 32.0% (29.4-34.7) overall; and 47.1% (42.6-51.5) for those with a DM-free interval >5 years. Conclusions Women’s risk of DM improves over time since diagnosis; and post-metastasis survival is longer for women with later DM. Key messages Health record linkage methods can be used for conditional risk estimates to inform women who remain DM-free after early BC about their risk of DM in subsequent years; and post-metastasis survival.