Prognostic relevance of the neurological symptom burden in brain metastases from breast cancer

Patient’s characteristics

A total of 6151 patients with BM from different solid tumours are included in the Vienna Brain Metastasis Registry. In total, 716/6151 (11.6%) patients with BM from BC and available information on the symptomatic status at diagnosis of BM were identified and included in this analysis (Supplementary Fig. 1). In 653/716 (91.2%) patients information on the receptor status was available including 260/653 (39.8%) patients with HR+ BC, 230/653 (35.2%) with HER2+ BC and 163/653 (25.0%) with TN-BC.

In total, 100/716 (14.0%) patients were diagnosed with Stage IV BC at time of primary tumour diagnosis. Of these, 20.0% (20/100) were synchronously diagnosed with primary tumour and BM. In patients with subsequent diagnosed BM, the median BMFS was 49 months (range 2–443 months).

See Table 1 and Supplementary Table 3 for further information on the clinical characteristics of the total patient cohort.

Neurological symptom burden at diagnosis of BM

In total, 80% (573/716) of the patients presented with neurological symptoms at the time of BM diagnosis (Fig. 1a). A numerical increase of asymptomatic patients at the time of BM diagnosis was observed over the last three decades (1992–1999: 4.8% vs. 2010–2020: 21.6%; p = 0.135; Table 2; Fig. 1b)—irrespective of the underlying BC subtype.

Presence of the neurological symptoms at diagnosis of BM (a + e) according to the year of BM diagnosis (b), BMFS (c), and subtypes of BC (d).

Significant differences according to symptomatic burden at BM diagnosis were observed between BC subtypes. Patients with HER2+ BC were more frequently diagnosed with asymptomatic BM during staging procedures (65/230;28.3%), compared to HR+ BC (47/260; 18.1%) and TN-BC (24/163; 14.7%; chi-square test; p = 0.002; Supplementary Table 2; Fig. 1d).

Focal deficits

Focal deficits were the most commonly documented neurological symptoms at the time of BM diagnosis (88.0%; 504/573; Fig. 1e). According to BC subtypes, patients with TN-BC (133/139; 95.7%) presented most frequently with focal deficits compared to HR+ BC (187/213; 87.8%) and HER2+ BC (136/165; 82.4%; p = 0.005; chi-square test; Supplementary Table 3).

Signs of increased intracranial pressure

In total, 275/573 (48.0%) patients showed signs of increased intracranial pressure at the time of BM diagnosis (Fig. 1e). No differences were observed according to BC subtypes (p > 0.05; chi-square test; Supplementary Table 3).

Epileptic seizures

In total, 15.7% (90/573) of symptomatic patients had an epileptic event at BM diagnosis. Of these, 42.2% (38/90) presented with focal seizures, 46.7% (42/90) with generalised seizures, and 11.1% (10/90) with combined focal and generalised seizures at the time of BM diagnosis (Fig. 1e). No significant differences were observed according to BC subtypes regarding the frequency of seizures at baseline (p > 0.05; chi-square test; Supplementary Table 3).

Neuropsychological symptoms

Neuropsychological symptoms were present in 31.9% (183/573) of symptomatic patients at baseline. Cognitive impairment/dysfunction (151/183; 82.5%) was most frequently documented, followed by organic brain syndrome (32/183; 17.5%), irrespective of the underlying BC subtype (p > 0.05; chi-square test; Supplementary Table 3; Fig. 1e).

See additional information on symptoms at BM diagnosis in Table 1, Supplementary Table 3, and Fig. 1e.

Clinical characteristics at diagnosis of BM associated with neurological symptom burden

BMFS was not statistically significantly different between patients with asymptomatic and symptomatic BM (48 vs. 49 months; p = 0.720; log-rank test, Table 2, Fig. 1c).

Patients with BM as the only metastatic site had a significantly higher rate of symptomatic disease compared to patients with known extracranial metastasis (p = 0.019; chi-square test; Table 2; Fig. 2a). This effect was most pronounced in HER2+ BC, as 83.7% (41/49) of patients with absent extracranial metastases showed neurological symptoms at diagnosis of BM compared to 68.5% (124/181) of patients with concurrent BM and extracranial disease (p = 0.036; chi-square test; Supplementary Table 2).

Clinical characteristics at BM diagnosis of BM in symptomatic and asymptomatic patients (a); treatment strategies after BM diagnosis in HR+ BC (b), HER2+ BC (c) and TN-BC (d) patients according to symptomatic status.

Furthermore, the size of BM at diagnosis was also significantly associated with neurological symptoms. A total of 92.6% (162/175) of patients with BM ≥ 3 cm showed neurological symptoms, while only 7.4% (13/175) of patients with BM < 3 cm were symptomatic at BM diagnosis (p < 0.001; chi-square test; Table 2; Fig. 2a). In particular, 92.2% of (47/51) patients with HER2+ BC and BM ≥ 3 cm presented with neurological symptoms compared to 64.5% (98/152) with BM < 3 cm (p < 0.001; chi-square test; Supplementary Table 2).

In contrast, the number of BM or the localisation (hemispheres/lobes) as well as the age and KPS of the patients at diagnosis of BM were not associated with the neurological burden (p > 0.05; chi-square test; Table 2; Supplementary Table 2).

All additional information on the clinical characteristics associated with the symptomatic burden at BM diagnosis is given in Table 3, Supplementary Table 2, and Fig. 2a.

Treatment strategies after diagnosis of BM according to neurological symptom burden

Radiotherapy, encompassing whole brain radiation therapy (WBRT) and focal radiotherapy presented as the predominant therapeutic modality following the diagnosis of BM in BC patients across the decades (Supplementary Table 3). However, we observed noteworthy variations in initial treatment strategies upon the symptomatology and BC subtype. Specifically, among symptomatic patients with HR+ BC, WBRT was the predominant choice (50.2%; 107/213), surpassing focal radiotherapy alone (15.5%; 33/213; p < 0.001; chi-square test; Supplementary Table 2; Fig. 2d).

In total, 14.6% (31/213) of HR+ BC patients with neurological symptoms received combinational approaches after diagnosis of BM: 18/31 (58.1%) underwent WBRT along with focal radiotherapy, while 13/31 patients (41.9%) received additional radiation therapy after neurosurgery at the time of BM diagnosis.

Conversely, symptomatic HER2+ BC patients predominantly underwent focal radiotherapy (39.4%; 65/165) alone at diagnosis of BM. A smaller subset (7.3%; 12/165) underwent additional WBRT following focal radiotherapy. WBRT as the sole initial treatment was chosen in 27.3% (45/165) of patients (p = 0.040, chi-square test; see Supplementary Table 2; Fig. 2c). Additionally, 13.9% (23/165) of symptomatic HER2+ BC patients received further radiotherapy after undergoing initial neurosurgery at BM diagnosis.

In asymptomatic patients, regardless of the underlying BC subtype, SRS consistently emerged as the most frequently employed treatment modality over the decades (Supplementary Table 4). Moreover, a discernible upward trend in the utilisation of systemic therapy as an initial treatment approach after diagnosis of BM was noted in asymptomatic patients across different decades (1992–1999: 0% vs. 2010–2020: 5.6%; Supplementary Table 4).

In the cohort of TN-BC patients, no significant differences in treatment approaches were found based on neurological symptoms at the time of BM diagnosis (p > 0.05; chi-square test; Supplementary Table 2; Fig. 2a).

Furthermore, no variations in terms of intracranial recurrence patterns or the end-of-life period concerning neurological symptom status have been observed. However, patients diagnosed with BM in an asymptomatic state exhibited a significantly prolonged median duration to extracranial progression compared to symptomatic patients (6 vs. 5 months; p = 0.008; log-rank test, Table 2, and Supplementary Table 2). Find a comprehensive overview in Table 2, Supplementary Table 2, and Figs. 2a–3f on therapeutic modalities, patterns of intra- and extracranial progression, and the end-of-life period in accordance with the neurological symptom burden.

Overall survival from diagnosis of BM according to the neurological symptom status at BM diagnosis in all patients (a) as well as in HR+ BC (b), HER2+ BC (c), and TN-BC (d) patients.

Association of neurological symptoms with survival prognosis

Patients without neurological symptoms at baseline had a significantly prolonged median OS from diagnosis of BM compared to symptomatic patients (20 vs. 9 months, p < 0.001; log-rank test; Table 2; Fig. 3a). This observed effect persisted consistently across all BC subtypes (HR+ BC 29 vs. 9 months; HER2+ BC 24 vs. 12 months; TN 12 vs. 6 months; p < 0.05; log-rank test; Supplementary Table 2; Fig. 3b–d) and throughout the three decades (1992–1999: 14 vs. 9 months; 2000–2009: 19 vs. 8 months; 2010–2020: 27 vs. 9 months; p < 0.05; log-rank test, Supplementary Table 4).

Subsequently, we analysed the prognosis based on the type of neurological symptom. Here, patients with neuropsychological symptoms presented a significantly shorter median OS compared to patients with other neurological symptoms (12 vs. 5 months; p < 0.001; log-rank test; Supplementary Table 7). Specifically, the presence of cognitive dysfunction at the time of BM diagnosis was correlated with a substantially compromised median OS in contrast to patients presenting with other neurological symptoms (12 vs. 6 months; p < 0.001; log-rank test; Supplementary Table 7).

Multivariate survival analysis including the symptomatic burden and established prognostic scores

To assess the prognostic independence of neurological symptoms in relation to other prognostic factors, we incorporated parameters such as Breast-GPA, the size of BM at diagnosis, treatment strategies (including WBRT, SRS, neurosurgery, systemic therapy, BSC), and neurological symptoms into a multivariate model. In this analysis, the presence of neurological symptoms (HR: 1.6; 95% CI, 1.35–1.95; p < 0.001; Supplementary Table 6A), the treatment approach after BM diagnosis (HR: 1.1; 95% CI, 1.03–1.17; p = 0.04) and the Breast-GPA (HR: 1.4; 95% CI, 1.27–1.49; p < 0.001; Cox-regression model; Supplementary Table 6A) remained independently associated with prognosis. In addition, we performed an intern validation with the bootstrapping method to prove our analysis. In this, the results of the bootstrapping indicated that our Cox-regression model is stable as consistent hazard ratio estimates across different subsets of the data (Supplementary Table 6B).

Furthermore, the impact of neurological symptoms on survival prognosis was irrespective of the Breast-GPA classes, as asymptomatic patients in good (2.5–4.0 points) as well as in lower (0.0–2.0 points) prognostic classes showed an improved median OS compared to symptomatic patients (p < 0.05; log-rank test; Supplementary Table 6).