Effect of the timing of recurrence on survival among patients with breast cancer with brain metastasis.

Abstract

e14005 Background: Patients with Metastatic Breast Cancer (MBC) and brain metastasis (BM) have poor prognosis. There is a lack of data on the effects of timing of MBC diagnosis [de novo (d) vs recurred (r) MBC] and BM diagnosis [dBM vs later onset (lo) BM] on overall survival (OS). We conducted a natural history-type study to investigate these effects among MBC BM patients. Methods: The data for this retrospective single institute study was extracted from a database of MBC BM patients diagnosed with BM from Jan 2010 to June 2021. Definitions: dMBC= MBC at initial diagnosis (Dx); rMBC= MBC with prior history of early stage BC; dBM= BM diagnosed< 6 weeks (wk) from MBC Dx; loBM= BM diagnosed ≥ 6 wk from MBC Dx. The following endpoints were compared between dMBC vs rMBC: 1. interval from MBC to BM Dx (Interval MBC-BM) 2. OS from MBC Dx 3. OS from BM Dx. OS from BM Dx was compared between dBM vs loBM. Median endpoints were estimated using the Kaplan-Meier method and reported with 95% confidence intervals (CI). Cohorts were compared using the Logrank test and adjusted for other variables using Cox models. Results: Interval MBC-BM, OS from MBC Dx, and BM Dx were not statistically different between dMBC and rMBC groups in unadjusted analyses (See Table) and when adjusted in multivariate analyses. Median OS from BM Dx was significantly longer in dBM group vs loBM group (See Table). In multivariate model adjusting for variables at BM Dx (age, race, performance status, visceral disease, number of BM >1cm, leptomeningeal disease, BC receptor types), risk of death was higher in loBM group vs dBM group (Hazard Ratio, HR = 0.499; p=0.006). In dMBC group, 28.1% had dBM compared to 45.6% in the rMBC (p=0.083, Chi-square test). Subset analyses: Among rMBC group, median OS from BM Dx with dBM (n=42, OS=44.3 months (m), 95% CI 19.0-Not evaluable (NE)) was significantly longer compared to loBM (n=50, OS=13.7m (6.6-21.4)), p=0.002 (LogRank Test); but among dMBC group, there was no significant difference in OS from BM Dx with dBM (n=9, OS=32.4m (12.9-122.5)) vs. with loBM (n=23, OS= 17.8m (6.6-NE)), p=0.885. Conclusions: In a retrospective database of patients with BM from BC, time from MBC to BM diagnosis, OS from MBC diagnosis, and OS from BM diagnosis did not differ between those with recurrent vs de novo MBC. Interestingly, in unadjusted and adjusted analyses, OS from BM diagnosis was longer for those with de novo BM diagnosed at/around MBC diagnosis than those whose BM was later in onset. Earlier-onset BM in the MBC disease course could be more sensitive to cancer-directed therapies. Studies are needed to test this hypothesis.[Table: see text]