Breast Cancer Characteristics and Survival among Users versus Non-users of Raloxifene

Raloxifene reduces breast cancer (BrCA) incidence. However, it is unclear whether it also reduces mortality from BrCA. For raloxifene to reduce incidence but not mortality, BrCA survival would have to be worse for raloxifene users than non-users. SEER-Medicare was used to identify women with invasive BrCA diagnosed from 2007-2015 at age 65-89 who had prior Medicare Part D (prescription drug) enrollment; BrCA characteristics and survival were assessed among raloxifene regular users (≥ 180 days in past year) versus non-users. Logistic regression was used to assess cancer characteristics. Two methods utilizing proportional hazards models were employed to assess BrCA-specific survival. In Method 1, survival was assessed adjusting for demographics, mammography use and chronic conditions in the subset with Medicare fee-for-service enrollment. In Method 2, predicted survival as a function of BrCA characteristics was modeled in non-users and the model applied to users to predict survival. A total of 116317 raloxifene non-users and 1223 regular users were identified. Users were significantly more likely to have hormone receptor negative cancers, but less likely to have T2+, N1+ and metastatic disease. There were 10869 and 101 breast cancer deaths in non-users and regular users, respectively. The HR (users versus non-users) for breast cancer-specific survival in Method 1 was 0.94 (95% CI:0.73-1.22). In Method 2, predicted survival was higher in users than non-users (8-year survival 84.9% versus 83.4%). Breast cancer-specific survival in raloxifene users was not worse than in non-users, providing indirect evidence that raloxifene reduces breast cancer mortality.