Clinical data from 126 patients with BCBM who received RT to the brain were analyzed. Median follow-up was 44 months (95% CI 24.9–63.1). Patients’ characteristics are shown in Table 1. The median age at first RT was 57.3 years (interquartile range [IQR] 47.9–67.9) and median KPS at initial BM diagnosis was 80% (IQR 70–90). Patients had mostly one or two BM-directed RT treatment courses (median 1.0; IQR 1–2), with a maximum of five courses. Across all therapy sessions, 65.1% (n = 82/126) patients remained ‘non-multi metastatic’ (≤ 15 BM). The median was 3 BM in total during follow-up (IQR 2–7), while 19 patients (15.1%) had only a single BM for the whole follow-up period and 47 patients presented with a single brain metastasis at initial diagnosis of BCBM. In all, 28.8% had multiple BM (> 15 BM; n = 36/126), and 6% (n = 7/126) developed leptomeningeal disease. A total of 392 countable BCBM were analyzed, including 42 locally recurrent metastases.

The multidisciplinary treatment of BCBM with repeated courses of local RT leads to complex medical histories. Figure 1 illustrates an overview of RT courses after diagnosis of BCBM. The type of initial RT was approximately equally distributed, with 46% WBRT (n = 58/126) and 54% (n = 68/126) SRS/FSRT. Initial metastasis surgery was mainly performed before the first SRS/FSRT with 71% (n = 46) of all surgeries during the follow-up period at this point. Salvage RT due to first BM progression after initial WBRT was only by SRS/FSRT. A total of 22 patients received salvage WBRT after initial SRS/FSRT. Of note, 27.8% of patients (n = 35) received no WBRT during the follow-up and could be treated with one or more salvage SRS/FSRT courses after BCBM progression.

Overview of radiotherapy (RT) courses after initial diagnosis of breast cancer brain metastases (BCBM). SRS stereotactic radiosurgery, FSRT fractionated stereotactic radiotherapy, WBRT whole-brain radiotherapy

Fractionation of WBRT was mostly 10 × 3 Gy. For SRS, doses between 18 Gy and up to 24 Gy were used. FSRT was mostly delivered in 3 × 9 Gy, 5 × 6–7 Gy, and 7 × 5 Gy, with the use of the more fractionated concepts mainly for cavity-RT after metastasis resection. Figure 2 shows the relative use of the different RT techniques for the years 2013–2020. From 2013–2015, WBRT was the most common type of RT for BCBM with approximately 60% declining to about 30% in 2020. Since 2018, stereotactic approaches have become more frequent than WBRT and especially the use of FSRT increased from 5% in 2013 to 32% in 2020.

Type of local radiotherapy (RT) for breast cancer brain metastases (BCBM) per year (2013–2020). SRS stereotactic radiosurgery, FSRT fractionated stereotactic radiotherapy, WBRT whole-brain radiotherapy

At the time of the first BCBM diagnosis, 76% of patients presented with extracerebral metastases. Thus, only 26 patients did not receive systemic therapy within 4 weeks before or after their first intracranial RT. A total of 89 patients received some form of systemic therapy, with conventional chemotherapy being the most common (n = 52), followed by HER2-targeted therapies (n = 38) and endocrine therapy (n = 22). Trastuzumab was the most frequently used HER2-targeted therapy, while paclitaxel was the most commonly administered chemotherapy, followed by capecitabine.

The median time interval between initial BC diagnosis and diagnosis of BCBM in this cohort was 46 months (IQR 23–89). A total of 21 patients (16.6%) showed a late occurrence of BCBM (TI ≥ 10 years). A shorter TI of < 5 years (n = 80; 63.5%) was not associated with a worse OS.

BM from hormone receptor-positive BC were significantly associated with a longer TI between ID and diagnosis of BCBM, whereas triple-negative BC shows a tendency towards early occurrence of BM. Patients with BM from HER2-positive BC had a remarkably longer survival after the first occurrence of BM (Fig. 3).

Time interval between initial diagnosis, diagnosis of breast cancer brain metastases (BCBM) and death/censored analyzed according to biological subtypes. Hormone receptor-positive (HR): estrogen receptor-positive (ER) and ± progesterone receptor-positive (PR). Human epidermal growth factor receptor 2 (HER2)-positive: HER2-overexpression and HR-negative. Triple negative: HR-negative and HER2-negative. Other subtypes: All cases not included in the subtype-specific upper bar are summarized in the lower bar for comparison. P-values were calculated for the differences in the logarithmic mean time interval for each subgroup using a two-sided t‑test with the left p-values corresponding to the black bars and the right p-values to the grey bars

Of note, dissimilar to the other presented results, the iBMV was only calculated for the cohort with countable metachronous BM, defined as a minimum of 6 months between the initial diagnosis of breast cancer and the first diagnosis of BCBM, and ≤ 15 metastases countable on the patient’s MRI (see “Methods” section; n = 92, Table 1B). The median iBMV was 0.48 BM/year (IQR 0.21–1.22). Of these patients, 66 had 1–3 metastases, 17 had 4–10 metastases, and 9 had 11–14 metastases.

Risk classes by the iBMV index are not consistently defined and a threshold of two is proposed by some authors [12, 13], but as the reported iBMV for breast cancer is lower, we used the median for separation into two groups like initially published by Soike et al. [11]. In the Kaplan–Meier estimate an iBMV index above the median of 0.48 BM/year was significantly associated with a worse OS (p = 0.025; Fig. 4).

Kaplan–Meier survival curve for initial brain metastasis velocity (iBMV) ≤ or > the median of 0.48 brain metastases/year. No. Number

In the univariate Cox regression, iBMV above the median was significantly associated with a higher risk of death (p = 0.031; HR 1.92 [95% CI 1.06–3.46]) and an increase of the iBMV index by 1.0 BM/year was associated with a 1.17 increase in hazard ratio (HR; p = 0.011 [95% CI 1.04–1.31]; Table 2).

In comparison, TI between the ID and the diagnosis of BM and the number of BM at this time point, both had a HR of 0.99 and 1.09, respectively, in a bivariate Cox regression model. This indicates that neither the TI nor the number of BM alone has a substantial impact on OS compared to both combined in the iBMV index.

Median OS in the whole cohort was 15 months (95% CI 9.0–21.0). In univariate analysis, presence of extracranial metastases, HER2 status, KPS, number of BCBM, receiving WBRT and iBMV above the median of 0.48 BM/year were identified as significant variables affecting survival. Of note, the analysis of iBMV was done in a cohort of 92 patients as described before, and results are therefore shown in separate columns per analyzed cohort. For this subcohort, only HER2 status, KPS, receiving WBRT, and iBMV had an association with OS in univariate analysis (Table 2).

To test whether the iBMV has a significant impact on OS, we analyzed all covariables in the iBMV cohort (status of all covariables available in n = 89 patients) and compared results to the multivariable analysis of the other covariables in the whole cohort (status of all covariables available in n = 118 patients). As per multivariable analysis in the iBMV cohort, iBMV (per increase by 1 BM/year) was significantly associated with worse OS (HR = 1.21; 95% CI 1.04–1.41; p = 0.012). In the iBMV cohort and the whole cohort, HER2-negative BC (HR = 2.26; 95% CI 1.34–3.84; p = 0.002) and KPS < 80% (HR = 2.60; 95% CI 1.60–4.22; p = < 0.001) were significantly associated with worse OS (Table 2).

An analysis of the impact of systemic therapies (chemotherapy, HER2-targeted therapy, and endocrine therapy) administered within 4 weeks before or after the first intracranial RT on OS revealed that the prognostic value of HER2 status depends on the receipt of HER2-targeted therapy. When these covariates are included in the multivariate analysis, HER2 status was not significantly associated with OS, although the trend of other variables remained unaffected. Notably, the iBMV was still significantly associated with worse OS in the iBMV cohort (HR = 1.26, 95% CI 1.10–1.49; p = 0.007) in this model.