Surgical outcomes and prognostic factors in 228 patients with brain metastases: a focus on en bloc resection and multimodal therapy Authors: Peng Sun, Chunyu Gu, Haoran Wang, Yanzhen Zhang, Jianping Cao, Junpeng Ji, Mingshan Zhang Published: 2025 ABSTRACT ObjectivesThis study aimed to evaluate the survival outcomes and identify prognostic factors in patients undergoing surgical resection for brain metastases (BMs), with a particular focus on the impact of surgical techniques and multimodal therapy sequences.MethodsWe conducted a retrospective analysis of 228 consecutive patients who underwent neurosurgical resection for BMs at a single institution (2012–2020). Primary endpoints were median overall survival (mOS) and survival rates, analyzed via Kaplan-Meier methodology. Prognostic factors were assessed using univariate and multivariate Cox proportional hazards models, with subgroup analyses based on primary tumor origin.ResultsThe cohort’s overall mOS was 14 months, with 1-, 2-, and 3-year survival rates of 60.3%, 35.4%, and 23.2%, respectively. Multivariate analysis identified preoperative Karnofsky Performance Status (KPS) as a powerful independent prognostic factor (Hazard Ratio [HR] = 49.076, 95% Confidence Interval [CI]: 21.483–112.110, P < 0.001). The treatment modality significantly influenced survival: patients receiving postoperative radiotherapy achieved an mOS of 18 months, compared to 13 months for surgery alone. Strikingly, the cohort receiving combined pre- and postoperative radiotherapy (n = 6) demonstrated a superior mOS of 41 months. In lung cancer BMs (n = 109, mOS = 11 months), targeted therapy was associated with a significantly prolonged mOS of 28 months versus 10 months in non-recipients. For breast cancer BMs (n = 26, mOS = 17 months), chemotherapy was an independent favorable prognostic factor (HR = 6.297, P = 0.009).ConclusionsSurgical resection, particularly when integrated into a multimodal strategy, confers significant survival benefits for patients with BMs. The sequence of therapy is critical, with combined perioperative radiotherapy emerging as a highly promising regimen. Preoperative KPS is a paramount prognosticator. These findings advocate for an aggressive, personalized surgical approach embedded within a comprehensive neuro-oncological framework.Keywords: Brain metastases, Surgical treatment, Early diagnosis, Survival time Introduction Brain metastases represent the most common intracranial malignancies in adults, posing a growing clinical challenge in an aging population and in the era of advanced systemic therapies where cancer patients live long enough to develop central nervous system involvement [1, 2]. Current epidemiological studies indicate that 2–14% of patients present with brain metastases prior to the diagnosis of their primary tumor [3]. The most common origins are lung cancer, breast cancer, and melanoma, with a significant proportion (40–60%) of patients presenting with multiple intracranial lesions at diagnosis [4, 5]. Moreover, tumors from different sources tend to have different metastasis sites when they develop intracranial metastasis, such as lung cancer tends to metastasize to the frontal, occipital and parietal lobes, while breast cancer tends to metastasize to the cerebellum more often(Fig. 1) [6, 7]. The management of brain metastases has evolved into a multimodal paradigm, wherein modern microneurosurgery serves as a cornerstone. This role is justified by evidence demonstrating that gross total resection is critical for local control and survival [8], and is further enhanced by integration with targeted therapies and radiotherapy. This study aims to systematically evaluate postoperative outcomes, quantify the survival benefit of combining surgery with adjuvant therapies, and establish tumor-specific algorithms to optimize individualized management. Metastases in superficial versus deep brain regions of malignancies. Representative neuroimaging findings: A Superficial metastasis (frontal lobe melanoma) and B deep-seated metastasis (basal ganglia breast cancer), both demonstrating irregular contrast-enhancing masses with extensive peritumoral edema, significant ventricular compression, and marked midline shift Materials and methods Study design and patient selection We conducted a single-center, retrospective cohort study of 228 consecutive patients who underwent surgical resection for brain metastases at Sanbo Brain Hospital, Capital Medical University, between January 2012 and June 2020. The study was approved by the Institutional Ethics Committee, and informed consent was obtained from all participants. Inclusion criteria were: (1) histopathological confirmation of brain metastasis from surgical specimens; (2) availability of pre-operative contrast-enhanced MRI; and (3) complete clinical and follow-up data. Patients with missing key data or who underwent only biopsy were excluded. Data collection Demographic, clinical, and tumor-related variables were systematically collected from electronic medical records. This included age, sex, Karnofsky Performance Status (KPS) at presentation, primary tumor type, number of intracranial metastases (categorized as solitary or multiple), tumor location (supratentorial, infratentorial, or both), and maximum tumor diameter. Treatment-related variables encompassed the extent of resection (gross total or subtotal), surgical approach, receipt and timing of radiotherapy (preoperative, postoperative, or combined), and administration of systemic therapy (chemotherapy or targeted agents). Surgical and adjuvant treatment The primary surgical goal was maximal safe resection. The decision to pursue en bloc versus piecemeal resection was at the operating surgeon’s discretion, based on tumor characteristics (size, consistency, and adjacency to eloquent areas). All patients were discussed in a multidisciplinary neuro-oncology tumor board. Adjuvant radiotherapy (either whole-brain radiotherapy or stereotactic radiosurgery) and systemic therapy were administered per institutional guidelines and contemporary standards of care. Statistical analysis Statistical analyses were performed using SPSS Statistics. Continuous variables were presented as median with interquartile range and compared using the Mann-Whitney U test. Categorical variables were expressed as frequencies and percentages and compared using the Chi-square or Fisher’s exact test, as appropriate. Overall survival was defined as the time from surgery to death from any cause and was analyzed using the Kaplan-Meier method, with comparisons between groups made by the log-rank test. Univariate and multivariate analyses were performed using Cox proportional hazards models to identify independent prognostic factors. A two-sided p-value of < 0.05 was considered statistically significant. Results Patient demographics and clinical characteristics A total of 228 patients (133 males, 58.3%; 95 females, 41.7%) who underwent surgical resection for brain metastases were included in this analysis. The baseline characteristics of the entire cohort, stratified by treatment modality, are detailed in Table 1. The most common primary tumor was lung cancer (109 patients, 47.8%), followed by breast cancer (26, 11.4%), gastrointestinal malignancies (23, 10.1%), and melanoma (22, 9.6%). The majority of patients (153, 67.1%) presented with solitary brain metastases. A comparison of baseline characteristics between survivors and non-survivors is provided in Table 2. Table 1General data of patientsClinical dataOperation only (n = 107)Radiotherapy before surgery(n = 32)Surgery before radiotherapy (n = 83)Preoperative and postoperative radiotherapy (n = 6)Total(n = 228)Sex Male6616483133 Female411635395Age <657531676179 ≥ 6532116049KPS Score <701247023 ≥ 709528766205Tumor depth Superficial metastasis5928705162 Deep-seated metastasis2515031 Both exist simultaneously2338135Maximum tumor diameter(cm) <331925267 ≥ 37623584161Number of brain metastases 17716553151 215712135 ≥ 315916242Tumor location Supratentorial7023555151 Subtentorial23416043 Both exist simultaneously14514134Targeted therapy Yes17617242 No9026664186Chemotherapy Yes26939579 No8123441149Degree of tumor resection Total resection7623644167 Subtotal resection22213138 Other residual lesions remain following gross total resection of the tumor976123Recurrent brain metastases Yes15824249 No9224594179 General data of patients Table 2Baseline characteristics by outcome groupsClinical characteristics (N = 228)Survivors (N = 82)Non-survivors (N = 146)P ValueSex0.211 Male48 (21.1)84 (36.8) Female34 (14.9)62 (27.2)Age0.306 <6571 (31.1)108 (47.4) ≥ 6511 (4.8)38 (16.7)KPS Score <0.001 <701 (0.4)22 (9.6) ≥ 7081 (35.5)124 (54.4)Tumor depth0.842 Superficial metastasis66 (28.9)114 (50.0) Deep-seated metastasis9 (3.9)16 (7.0) Both exist simultaneously7 (3.1)16 (7.0)Maximum tumor diameter(cm)0.222 <319 (8.3)27 (11.8) ≥ 363 (27.6)119 (52.2)Number of brain metastases(N = 227)0.376 160 (26.4)92 (40.5) 210 (4.4)24 (10.6) ≥ 311 (4.8)30 (13.2)Tumor location0.909 Supratentorial57 (25.1)96 (42.3) Subtentorial16 (7.0)25 (11.0) Both exist simultaneously9 (4.0)24 (10.6)Targeted therapy 0.015 Yes30 (13.2)13 (5.7) No52 (22.8)133 (58.3)Chemotherapy0.079 Yes42 (18.4)40 (17.5) No40 (17.5)106 (46.5)Radiotherapy 0.044 No35 (15.4)84 (36.8) Preoperative8 (3.5)18 (7.9) Postoperative36 (15.8)41 (18.0) Preoperative and postoperative3 (1.3)3 (1.3)Degree of tumor resection(N = 224)0.226 Total resection65 (28.5)102 (44.7) Subtotal resection12 (5.3)26 (11.4) Other residual lesions remain following gross total resection of the tumor5 (2.2)18 (7.9)Recurrent brain metastases0.129 Yes20 (8.8)32 (14.0) No62 (27.2)114 (50.0)Continuous variables are presented as median (interquartile range, IQR), while categorical variables are expressed as frequency (percentage, %). The Mann-Whitney U test was used to determine P-values for continuous variables, with Pearson’s chi-square or continuity-corrected chi-square tests applied for categorical variables.Bold indicates meaningful variables Baseline characteristics by outcome groups Continuous variables are presented as median (interquartile range, IQR), while categorical variables are expressed as frequency (percentage, %). The Mann-Whitney U test was used to determine P-values for continuous variables, with Pearson’s chi-square or continuity-corrected chi-square tests applied for categorical variables.Bold indicates meaningful variables Overall survival and prognostic factors for the entire cohort At 2 weeks postoperation, neurological symptoms had completely resolved in 103 cases (45.2%), significantly improved in 102 (44.7%), and showed no improvement or worsened in 13 (5.7%). Perioperative mortality occurred in 10 cases (4.4%). During the 6-month follow-up period, 31 deaths occurred, including 1 surgery-related mortality and 30 cases attributed to systemic complications from end-stage widespread metastatic disease. Long-term follow-up demonstrated survival rates of 60.3% at 1 year, 35.4% at 2 years, 23.2% at 3 years, and 13.7% at 5 years, with a median overall survival of 14 months (Tables 3 and 4). The median overall survival (mOS) for the entire cohort was 14 months, with 1-year, 2-year, and 3-year survival rates of 60.3%, 35.4%, and 23.2%, respectively (Table 4). Univariable analysis identified preoperative KPS score, chemotherapy, and radiotherapy as significant factors influencing survival (Table 5). In the subsequent multivariable Cox regression analysis, only the preoperative KPS score retained independent prognostic significance (Hazard Ratio [HR] = 49.076, 95% Confidence Interval [CI]: 21.483–112.110, P < 0.001). Patients with a KPS score ≥ 70 had a markedly superior mOS of 17 months, compared to 3 months for those with a KPS score < 70 (Table 3). Table 3Survival rates and median overall survival (mOS)1-year survival rate (%)2-year survival rate (%)3-year survival rate (%)mOS (months)Sex Male59.930.118.314 Female60.741.122.417Age <6563.837.824.017 ≥ 6548.527.811.912KPS Score <700003 ≥ 7066.839.225.717Tumor depth Superficial metastasis60.534.624.415 Deep-seated metastasis49.428.814.412 Both exist simultaneously71.639.013.014Maximum tumor diameter(cm) <361.136.020.618 ≥ 360.135.222.514Number of brain metastases 159.437.027.514 256.725.519.613 ≥ 364.837.915.114Tumor location Supratentorial60.037.725.815 Subtentorial62.135.325.715 Both exist simultaneously58.024.89.913Chemotherapy Yes82.654.836.625 No48.525.316.112Radiotherapy No53.930.413.713 Preoperative46.730.318.213 Postoperative65.041.033.618 Preoperative and postoperative80.062.553.341Degree of tumor resection Total resection60.637.225.915 Subtotal resection58.626.48.813 Other residual lesions remain following gross total resection of the tumor60.935.815.318Recurrent brain metastases Yes66.942.030.019 No65.033.521.014 Survival rates and median overall survival (mOS) Table 4Comparative analysis of survival outcomes in brain metastasis patientsTumor origin1-year survival rate (%)2-year survival rate (%)3-year survival rate (%)5-year survival rate (%)mOS (months)Above60.335.423.213.714Lung48.333.226.916.811Breast54.626.714.23.817Gastrointestinal origin31.44.50010Malignant melanoma17.75.94.808 Comparative analysis of survival outcomes in brain metastasis patients Table 5Univariable and multivariable Cox regression analyses of survival-associated variablesVariablesUnivariable analysisMultivariable analysisP ValueHR95% CIP ValueHR95% CISex0.1210.7680.550–1.0720.8800.9730.680–1.391Age0.0531.4420.996–2.0890.5101.1380.774–1.673KPS Score <0.001 53.38624.524–116.218–20.108 <0.001 49.07621.483–112.110Maximum tumor diameter(cm)0.8371.0450.687–1.5900.0833.0010.365–1.064Chemotherapy <0.001 2.2791.571–3.3050.4550.5580.481–1.388Radiotherapy 0.030 0.957No1.0001.000Preoperative0.6680.8920.529–1.5040.8190.9370.539–1.630Postoperative 0.010 0.6090.417–0.8880.6220.9040.604–1.352Preoperative and Postoperative0.0720.3450.108 − 1.0990.7360.8070.233–2.798Degree of tumor resection0.4160.513–1.0850.275Total resection1.0001.000Subtotal resection0.1941.3310.864–2.0500.8550.0330.552–2.048Other residual lesions remain following gross total resection of the tumor0.6261.1330.685–1.8740.1062.6190.170–1.185Bold indicates meaningful variables Univariable and multivariable Cox regression analyses of survival-associated variables Bold indicates meaningful variables Impact of treatment modalities on survival The sequence of multimodal therapy demonstrated a profound impact on patient outcomes (Table 3). Patients treated with surgery alone had an mOS of 13 months. The addition of postoperative radiotherapy significantly extended the mOS to 18 months. Most notably, the small cohort of patients (n = 6) who received combined preoperative and postoperative radiotherapy achieved a remarkable mOS of 41 months. Furthermore, the administration of chemotherapy was associated with a significant survival benefit, yielding an mOS of 25 months versus 12 months for those not receiving chemotherapy (Table 3 , Figs.2-4). Table 6Baseline characteristics by postoperative outcomes in lung cancerClinical characteristics (N = 109)Survivors (N = 43)Non-survivors (N = 66)P ValueSex0.944 Male27 (24.8)41 (37.6) Female16 (1479)25 (22.9)Age 0.010 <6538 (34.9)44 (40.4) ≥ 655 (4.6)22 (20.2)KPS Score 0.003 <700 (0.0)12 (11.0) ≥ 7043 (39.4)54 (49.5)Tumor depth0.965 Superficial metastasis35 (32.1)55 (50.5) Deep-seated metastasis3 (2.8)4 (3.7)Maximum tumor diameter(cm)0.596 <311 (10.1)14 (12.8) ≥ 332 (29.4)52 (47.7)Number of brain metastases0.151 130 (27.5)35 (32.1) 27 (6.4)12 (11.0) ≥ 36 (5.5)19 (17.4)Tumor location0.388 Supratentorial24 (22.0)37 (33.9) Subtentorial13 (11.9)14 (12.8) Both exist simultaneously6 (5.5)15 (13.8)Pathological type0.182 Adenocarcinoma25 (22.9)48 (44.0) Squamous cell carcinoma3 (2.8)2 (1.8) Small cell carcinoma13 (11.9)12 (11.0) Large cell carcinoma2 (1.8)1 (0.9) Others0 (0.0)3 (2.8)Targeted therapy 0.038 Yes17 (15.6)14 (12.8) No26 (23.9)52 (47.7)Chemotherapy 0.031 Yes24 (22.0)23 (21.1) No19 (17.4)43 (39.4)Radiotherapy0.176 No13 (11.9)32 (29.4) Preoperative4 (3.7)8 (7.3) Postoperative23 (21.1)24 (22.0) Preoperative and postoperative3 (2.8)2 (1.8)Degree of tumor resection 0.044 Total resection34 (31.2)44 (40.4) Subtotal resection8 (7.3)10 (9.2) Other residual lesions remain following gross total resection of the tumor1 (0.9)12 (11.0)Recurrent brain metastases0.496 Yes12 (11.4)22 (21.0) No30 (28.6)41 (39.0)Continuous variables: median (IQR), compared via Mann-Whitney U test; categorical variables: n (%), analyzed with Pearson’s chi-square or continuity correction, as appropriateBold indicates meaningful variables Baseline characteristics by postoperative outcomes in lung cancer Continuous variables: median (IQR), compared via Mann-Whitney U test; categorical variables: n (%), analyzed with Pearson’s chi-square or continuity correction, as appropriate Bold indicates meaningful variables Survival outcomes stratified by preoperative KPS score Survival outcomes stratified by radiotherapy timing Subgroup analysis by primary tumor origin Lung cancer brain metastases (n = 109) For patients with lung cancer, the mOS was 11 months (Table 4 and 6). Multivariable analysis confirmed KPS score as a powerful independent prognostic factor (HR = 52.774, 95% CI: 16.732–166.456, P < 0.001; Table 7). A striking survival advantage was observed in patients receiving targeted therapy, with an mOS of 28 months compared to 10 months for non-recipients (Table 8 , Figs 5-10). Table 7Univariate and multivariate Cox analyses of survival predictors in lung cancerVariablesUnivariable analysisMultivariable analysisP ValueHR95% CIP ValueHR95% CISex0.1400.6800.408–1.1340.1480.6450.356–1.168Age 0.002 2.3261.379 − 3.9240.1001.7750.896–3.517KPS Score <0.001 42.38215.856–113.281 <0.001 52.77416.732–166.456Tumor depth0.9060.851Superficial metastasis1.0001.000Deep-seated metastasis0.8711.0880.393–3.0100.9760.9820.310–3.115Tumor depth0.6961.2780.373–4.3750.7221.3160.289–5.989Tumor location0.1940.807Supratentorial1.0001.000Subtentorial0.9770.9910.531–1.8480.7621.1270.520–2.442Both exist simultaneously0.0861.7170.927–3.1830.5441.2900.567–2.939Pathological type 0.018 0.281Adenocarcinoma1.0001.000Squamous cell carcinoma0.3990.5420.131–2.2490.9160.9230.205–4.156Small cell carcinoma0.2481.4740.763–2.8470.9350.9650.415–2.248Large cell carcinoma0.5230.5240.072–3.8070.0970.1640.019–1.389Others 0.002 6.9592.084–23.2320.1942.8010.592–13.264Targeted therapy 0.010 2.1901.203–3.9860.2311.5510.756– 3.182Yes 0.005 2.1191.259–3.5680.3081.4140.726–2.752No 0.037 0.204Chemotherapy1.0001.000Yes0.1280.5270.231–1.2020.0700.4120.158–1.075No 0.019 0.5280.310–0.8990.2560.6910.365–1.308Radiotherapy0.0580.2480.059–1.0460.2010.3510.070–1.748Degree of tumor resection0.4540.393–1.1610.760Total resection1.0001.000Subtotal resection0.5311.2490.624–2.5010.6251.2210.548–2.719Other residual lesions remain following gross total resection of the tumor0.2341.4800.776–2.8230.5161.3480.547–3.321CI: Confidence interval; HR: Hazard ratioBold indicates meaningful variables Univariate and multivariate Cox analyses of survival predictors in lung cancer CI: Confidence interval; HR: Hazard ratio Bold indicates meaningful variables Table 8Survival outcomes by prognostic factors in lung cancer brain metastases1-year survival rate (%)2-year survival rate (%)3-year survival rate (%)mOS (months)Sex Male40.125.918.410 Female56.342.429.120Age <6558.840.431.920 ≥ 6517.611.73.89KPS Score <700002 ≥ 7054.437.430.317Tumor depth Superficial metastasis47.531.526.311 Deep-seated metastasis50.050.025.010Maximum tumor diameter(cm) <336.218.18.814 ≥ 352.930.821.717Number of brain metastases 145.333.628.511 250.543.138.313 ≥ 333.322.215.611 Tumor depth42.712.26.111Tumor location Supratentorial50.236.027.713 Subtentorial50.644.329.517 Both exist simultaneously30.310.1010Pathological type Adenocarcinoma52.035.227.514 Squamous cell carcinoma50.050.050.011 Small cell carcinoma35.80010 Large cell carcinoma66.733.333.314 Others0002Targeted therapy Yes71.556.530.428 No37.223.517.610Chemotherapy Yes71.743.834.422 No32.126.37.910Radiotherapy No26.322.717.510 Preoperative64.325.7021 Postoperative57.033.322.217 Preoperative and postoperative75.075.025.028Degree of tumor resection Total resection48.642.334.712 Subtotal resection31.415.75.911 Other residual lesions remain following gross total resection of the tumor53.815.48.318Recurrent brain metastases Yes54.133.327.713 No45.134.828.911 Survival outcomes by prognostic factors in lung cancer brain metastases Age-stratified survival curves in lung cancer brain metastases Survival outcomes stratified by radiotherapy timing in lung cancer brain metastases Breast cancer brain metastases (n = 26) This subgroup exhibited one of the most favorable outcomes(Table 9), with an mOS of 17 months (Table 4). Multivariable analysis identified chemotherapy as an independent favorable prognostic factor (HR = 6.297, 95% CI: 1.594–24.868, P = 0.009; Table 10, Figs. 11-13). Table 9Clinicopathological features of breast cancer patients grouped by postoperative outcomesClinical characteristics (N = 26)Survivors (N = 7)Non-survivors (N = 19)P ValueAge0.790 <656 (23.1)17 (65.4) ≥ 651 (3.8)2 (7.7)KPS Score0.444 <701 (3.8)1 (3.8) ≥ 706 (23.1)18 (69.2)Tumor depth0.131 Superficial metastasis7 (26.9)14 (53.8) Deep-seated metastasis0 (0.0)0 (0.0) Both exist simultaneously0 (0.0)5 (19.2)Maximum tumor diameter(cm)(N = 24)0.384 <32 (8.3)3 (12.5) ≥ 34 (16.7)15 (62.5)Number of brain metastases0.121 16 (23.1)8 (30.8) 21 (3.8)6 (23.1) ≥ 30 (0.0)5 (19.2)Tumor location0.226 Supratentorial5 (19.2)12 (46.2) Subtentorial2 (7.7)2 (7.7) Both exist simultaneously0 (0.0)5 (19.2)Targeted therapy0.463 Yes2 (7.7)3 (11.5) No5 (19.2)15 (61.5)Chemotherapy0.117 Yes5 (19.2)7 (26.9) No2 (7.7)12 (46.2)Radiotherapy 0.025 No3 (11.5)14 (53.8) Preoperative0 (0.0)3 (11.5) Postoperative4 (15.4)1 (3.8) Preoperative and postoperative0 (0.0)1 (3.8)Degree of tumor resection0.449 Total resection6 (23.1)12 (46.2) Subtotal resection1 (3.8)12 (46.2) Other residual lesions remain following gross total resection of the tumor0 (0.0)3 (11.5)Recurrent brain metastases0.444 Yes1 (3.8)1 (3.8) No6 (23.1)18 (69.2)Continuous variables: median (IQR), compared via Mann-Whitney U test; categorical variables: n (%), analyzed with Pearson’s chi-square or continuity correction, as appropriateBold indicates meaningful variables Clinicopathological features of breast cancer patients grouped by postoperative outcomes Continuous variables: median (IQR), compared via Mann-Whitney U test; categorical variables: n (%), analyzed with Pearson’s chi-square or continuity correction, as appropriate Bold indicates meaningful variables Table 10Prognostic factors in breast cancer brain metastases: univariate and multivariate Cox analysesVariablesUnivariable analysisMultivariable analysisP ValueHR95% CIP ValueHR95% CIMaximum tumor diameter(cm)0.6870.7720.219–2.7170.9041.0890.270–4.389Number of brain metastases0.8710.21211.0001.00020.6051.3250.455–3.8560.2222.0880.640–6.811≥ 30.9021.0730.348–3.3090.3650.5290.134–2.094Targeted therapy0.4771.5710.453–5.4470.6971.3790.274–6.941Chemotherapy 0.010 4.0651.398–11.815 0.009 6.2971.594–24.868Radiotherapy0.1990.339No1.0001.000Preoperative0.8551.1270.311–4.0800.9171.0750.279–4.134Postoperative 0.035 0.1100.014–0.8550.1450.1980.023–1.744Preoperative and postoperative0.8221.2690 0.159– 10.0960.3080.2490.029–3.483CI: Confidence interval; HR: Hazard ratioBold indicates meaningful variables Prognostic factors in breast cancer brain metastases: univariate and multivariate Cox analyses CI: Confidence interval; HR: Hazard ratio Bold indicates meaningful variables Survival outcomes stratified by targeted therapy in breast cancer brain metastases Survival outcomes stratified by radiotherapy timing in breast cancer brain metastases Gastrointestinal cancer brain metastases (n = 23) Patients with gastrointestinal primaries had an mOS of 10 months, and the 3-year survival rate was 0% (Table 4). While univariable analysis suggested several negative prognostic factors (KPS < 70, ≥ 3 metastases, gastric origin, incomplete resection), none retained significance in the multivariable model (Tables 11 and 12, Figs. 14-17). Table 11Clinicodemographic features by outcome groups in resected GI cancer brain metastasesClinical characteristics (N = 23)survivors (N = 6)Non-survivors (N = 17)P ValueSex0.059 Male6 (26.1)10 (43.5) Female0 (0.0)7 (30.4)Age0.059 <656 (26.1)10 (43.5) ≥ 650 (0.0)7 (30.4)KPS Score0.544 <700 (0.0)1 (4.3) ≥ 706 (26.1)16 (69.6)Tumor depth0.544 Superficial metastasis6 (26.1)16 (69.6) Deep-seated metastasis0 (0.0)1 (4.3) Both exist simultaneously0 (0.0)0 (0.0)Maximum tumor diameter(cm)(N = 21)0.844 <31 (4.8)2 (9.5) ≥ 35 (23.8)13 (61.9)Number of brain metastases0.668 15 (21.7)12 (52.2) 21 (4.3)3 (13.0) ≥ 30 (0.0)2 (8.7)Tumor location0.527 Supratentorial5 (21.7)11 (47.8) Subtentorial0 (0.0)3 (13.0) Both exist simultaneously1 (4.3)3 (13.0)Primary tumor site0.547 Esophagus2 (8.7)2 (8.7) Stomach1 (4.3)2 (8.7) Colon2 (8.7)11 (47.8) Rectum1 (4.3)2 (8.7)Chemotherapy0.638 Yes2 (8.7)4 (17.4) No4 (17.4)13 (56.5)Radiotherapy0.923 No3 (13.0)8 (34.8) Preoperative1 (4.3)2 (8.7) Postoperative2 (8.7)7 (30.4) Preoperative and postoperative0 (0.0)0 (0.0)Degree of tumor resection0.426 Total resection6 (26.1)13 (56.5) Subtotal resection0 (0.0)3 (13.0) Other residual lesions remain following gross total resection of the tumor0 (0.0)1 (4.3)Recurrent brain metastases0.231 Yes2 (8.7)2 (8.7) No4 (17.4)15 (65.2)Continuous variables: median (IQR), compared via Mann-Whitney U test; categorical variables: n (%), analyzed with Pearson’s chi-square or continuity correction, as appropriate Clinicodemographic features by outcome groups in resected GI cancer brain metastases Continuous variables: median (IQR), compared via Mann-Whitney U test; categorical variables: n (%), analyzed with Pearson’s chi-square or continuity correction, as appropriate Table 12Prognostic factors in GI cancer brain metastases: from univariate screening to multivariable Cox modelVariablesUnivariable analysisMultivariable analysisP ValueHR95% CIP ValueHR95% CIKPS Score 0.030 21.4941.344-343.7030.16876.3410.161–36.271Number of brain metastases 0.033 0.48111.0001.00020.8021.1780.328–4.2340.2853.7980.329–43.855≥ 3 0.009 14.3621.948-105.8840.6260.2830.002–45.563Primary tumor site0.1290.204Esophagus1.0001.000Stomach 0.028 18.1361.378-238.6340.11354.8700.385–78.121Colon0.0972.84647.0990.13826.2740.349–19.777Rectum0.4702.4250.219–26.8760.5604.0860.036–463.100Radiotherapy0.4690.445No1.0001.000Preoperative0.4810.5700.119–2.7230.2130.1790.0412–2.687Postoperative0.2440.5210.174–1.5590.4230.5660.141–2.272Preoperative and postoperative––– – ––Degree of tumor resection0.0900.275Total resection1.0001.000Subtotal resection0.7341.2990.287–5.8760.27510.5480.154–724.595Other residual lesions remain following gross total resection of the tumor 0.029 22.2381.379–358.7210.16876.3410.161–36.271CI: Confidence interval; HR: Hazard ratioBold indicates meaningful variables Prognostic factors in GI cancer brain metastases: from univariate screening to multivariable Cox model CI: Confidence interval; HR: Hazard ratio Bold indicates meaningful variables Survival outcomes stratified by KPS in gastrointestinal tract tumors Survival outcomes stratified by radiotherapy in gastrointestinal tract tumors Melanoma brain metastases (n = 22) This subgroup had the poorest prognosis, with an mOS of 8 months (Table 4). Consistent with the overall cohort, KPS score was the sole independent prognostic factor in multivariable analysis (HR = 46.015, 95% CI: 4.345–487.356, P = 0.001; Tables 13 and 14, Figs. 18 and 19). Table 13Clinicopathologic features by outcome groups in malignant melanoma brain metastasesClinical characteristics (N = 22)survivors (N = 4)Non-survivors (N = 18)P ValueSex0.095 Male4 (18.2)10 (45.5) Female0 (0.0)8 (36.8)Age0.629 <654 (18.2)17 (77.3) ≥ 650 (0.0)1 (4.5)KPS Score0.176 <700 (0.0)6 (27.3) ≥ 704 (18.2)12 (54.5)Tumor depth0.056 Superficial metastasis2 (9.1)14 (63.6) Deep-seated metastasis2 (9.1)1 (4.5) Both exist simultaneously0 (0.0)3 (13.6)Maximum tumor diameter (cm)0.259 <32 (9.1)4 (18.2) ≥ 32 (9.1)14 (63.6)Number of brain metastases0.297 14 (18.2)14 (63.6) 20 (0.0)0 (0.0) ≥ 30 (0.0)4 (18.2)Tumor location0.059 Supratentorial2 (9.1)15 (68.2) Subtentorial2 (9.1)1 (4.5) Both exist simultaneously0 (0.0)2 (9.1)Chemotherapy0.259 Yes2 (9.1)4 (18.2) No2 (9.1)14 (63.6)Radiotherapy0.678 No3 (13.6)11 (50.0) Preoperative1 (4.5)4 (18.2) Postoperative0 (0.0)3 (13.6) Preoperative and postoperative0 (0.0)0 (0.0)Degree of tumor resection0.634 Total resection3 (13.6)14 (63.6) Subtotal resection1 (4.5)2 (9.1) Other residual lesions remain following gross total resection of the tumor0 (0.0)2 (9.1)Recurrent brain metastases0.910 Yes1 (4.5)5 (22.7) No3 (13.6)13 (59.1)Continuous variables: median (IQR), compared via Mann-Whitney U test; categorical variables: n (%), analyzed with Pearson’s chi-square or continuity correction, as appropriate Clinicopathologic features by outcome groups in malignant melanoma brain metastases Continuous variables: median (IQR), compared via Mann-Whitney U test; categorical variables: n (%), analyzed with Pearson’s chi-square or continuity correction, as appropriate Table 14Prognostic factors in melanoma brain metastases: univariate screening and multivariate Cox modelingVariablesUnivariable analysisMultivariable analysisP ValueHR95% CIP ValueHR95% CIKPS Score 0.001 40.6234.630-356.417 0.001 46.0154.345-487.356Number of brain metastases 0.016 4.4401.315–14.9970.1415.4950.567–53.225Chemotherapy0.2491.3320.022–2.6970.6441.4470.302–6.930Radiotherapy0.5000.786No1.0001.000Preoperative0.8550.0330.113–13.7850.5001.7380.348–8.682Postoperative0.9820.0010.013–15.7160.7090.6890.098–4.849Preoperative and postoperative––––––Degree of tumor resection0.2600.513–1.0850.773Total resection1.0001.000Subtotal resection0.2581.2700.360–43.9770.9661.0370.193–5.585Other residual lesions remain following gross total resection of the tumor0.2131.5520.246–9.6340.4730.2440.005– 11.522CI: Confidence interval; HR: Hazard ratioBold indicates meaningful variables Prognostic factors in melanoma brain metastases: univariate screening and multivariate Cox modeling CI: Confidence interval; HR: Hazard ratio Bold indicates meaningful variables Survival outcomes stratified of KPS in malignant melanoma Survival outcomes stratified of number of brain metastases in malignant melanoma Postoperative neurological outcomes Surgical intervention resulted in significant neurological improvement. At two weeks post-operation, neurological symptoms had completely resolved in 103 patients (45.2%) and significantly improved in 102 patients (44.7%). Among patients surviving beyond 3 years, 75.9% (22/29) had solitary brain metastases while only 24.1% (7/29) presented with multiple intracranial lesions, suggesting solitary metastasis may serve as a favorable prognostic indicator compared to multiple metastases. Importantly, early diagnosis and intervention play critical roles in managing cerebral metastases, with surgical resection demonstrating significant benefits in both prolonging survival and improving quality of life for affected patients. Discussion Our large, single-center retrospective study provides robust evidence supporting an aggressive surgical approach for brain metastases. The observed median overall survival (mOS) of 14 months for the entire cohort, surpassing the historical benchmarks of 1 month for untreated patients and 4–6 months for those receiving whole-brain radiotherapy alone, underscores the pivotal role of cytoreductive surgery in the modern management of this disease [9]. And relevant studies have also shown that patients undergoing multimodal therapy (surgery combined with postoperative chemoradiotherapy) demonstrate extended median survival of 10–15 months [10–12]. More importantly, our data transcend the mere demonstration of surgical efficacy; they delineate a hierarchy of prognostic factors and reveal profound insights into the optimization of multimodal therapy sequences. The overarching findings highlight the paramount importance of patient selection based on performance status, the critical integration of systemic therapies, and the potentially transformative benefit of a consolidated perioperative radiotherapy strategy. The paramount importance of patient selection and KPS The most robust finding across all our analyses was the overwhelming prognostic dominance of the preoperative KPS score. Its status as the sole independent prognostic factor in the overall cohort and in several primary-specific subgroups (e.g., lung cancer, melanoma) reinforces its indispensable role in patient selection [13, 14]. The stark contrast in mOS between patients with KPS ≥ 70 (17 months) and < 70 (3 months) provides a quantitative basis for clinical decision-making. This evidence strongly suggests that surgical intervention yields the greatest return in patients with preserved functional status, aligning with the core principles of graded prognostic assessment , 3 . Survival outcomes stratified by chemotherapy Survival outcomes stratified by KPS in lung cancer brain metastases Survival curves by histologic subtypes in lung cancer brain metastases Survival outcomes stratified by targeted therapy in lung cancer brain metastases Survival outcomes stratified by chemotherapy in lung cancer brain metastases Survival outcomes stratified by chemotherapy in breast cancer brain metastases Survival outcomes stratified by number of brain metastases in gastrointestinal tract tumors Survival outcomes stratified by degree of tumor resection in gastrointestinal tract tumors The efficacy of an aggressive surgical strategy and technique Our cohort’s favorable survival outcomes, particularly in traditionally poor-prognosis groups like melanoma (mOS 8 months) and gastrointestinal cancers (mOS 10 months), lend robust support to an expanded and aggressive surgical strategy. The fact that survival did not significantly differ between patients with single and multiple metastases in our cohort challenges a historical contraindication and suggests that in selected patients, maximal cytoreduction of all detectable lesions can be beneficial [15]. This approach is further justified by the immediate symptomatic relief achieved, with 89.9% of patients showing neurological improvement, which is crucial for preserving quality of life and enabling subsequent therapies. Regarding surgical technique, while our study was not designed to directly compare en bloc and piecemeal resection, our institutional emphasis on en bloc removal whenever feasible is supported by the literature. Our favorable local control outcomes are consistent with the work of Patel et al., who demonstrated that en bloc resection reduces local recurrence without increasing neurological morbidity [16, 17]. This technique theoretically minimizes the risk of spillage and cerebrospinal fluid dissemination, a principle that underpins our surgical philosophy. The multimodal paradigm: radiotherapy sequencing and systemic therapy Perhaps the most provocative finding of our study pertains to the sequencing of radiotherapy. While postoperative SRS to the resection cavity is the established standard, our data reveal a striking association between combined exceptional survival (mOS 41 months) and preoperative and postoperative radiotherapy. Although this group was small (n = 6) and likely reflects a selection bias towards patients with favorable disease biology and performance status, the magnitude of the survival difference is noteworthy. It generates the hypothesis that intensifying local control through a “sandwich” radiotherapy approach—preoperatively to sterilize the tumor periphery and facilitate resection, and postoperatively to eradicate residual microscopic disease—may yield synergistic benefits. This concept warrants further investigation in prospective studies. The critical importance of integrating effective systemic therapy is unequivocally demonstrated in our subgroup analyses. The near-tripling of mOS in lung cancer patients receiving targeted therapy (28 vs. 10 months) underscores a modern paradigm shift: surgery creates a window of opportunity for molecularly-guided systemic treatments to exert their effect on both intracranial and extracranial disease [18]. Existing literature reports a median survival of approximately 1 month for untreated brain metastasis patients [19]. Similarly, the identification of chemotherapy as an independent favorable factor in breast cancer brain metastases highlights the necessity of a coordinated, multidisciplinary effort where surgery is one component of a comprehensive treatment plan [20]. Early treatment, surgical approaches, and adjunctive technologies for brain metastases For patients with intracranial metastases, early diagnosis and appropriate treatment selection significantly improve survival and quality of life [21]. Complete surgical resection remains the strongest predictor of favorable survival and neurological outcomes [22, 23], and is indicated for solitary lesions or up to three metastases [15]. It provides rapid relief of mass effect and edema [24], and can be effectively combined with SRS [25]. Specific scenarios guide modality selection: large (>3 cm) symptomatic lesions with edema warrant resection, often followed by WBRT; smaller (≤ 3 cm) lesions with minimal mass effect may be managed with FSRT and WBRT. Although surgery with WBRT shows superior outcomes to surgery alone, its optimality is debated [9, 26]. Prognosis is independently influenced by age, KPS, systemic disease status, tumor histology, and intracranial tumor burden [27]. The principal advantages of surgery include obtaining diagnostic tissue (10–15% of presumed metastases are other pathologies), immediate symptom relief, potential local cure, and seizure control. In multiple metastases, resecting the dominant symptomatic lesion is prioritized [28], as is surgery for radio/chemotherapy-resistant cystic metastases or hemorrhagic lesions requiring urgent intervention. In terms of surgical approach, complete resection is particularly crucial when removing intracranial metastatic lesions. Yoo et al. conducted a prospective study utilizing microscopic gross total resection technique for intracranial metastases, where after achieving macroscopic complete tumor removal, an additional 5 mm margin of surrounding normal brain tissue was excised. The results demonstrated a significantly lower 1-year local recurrence rate of 29% in patients undergoing this surgical method compared to those receiving near-total resection (P < 0.01) [29]. However, another study found that the extent of resection for BMs demonstrated no significant impact on either local recurrence (LR) or OS [30]. Beyond striving for complete resection, en bloc removal should also be prioritized when feasible. In 2010, Patel et al. showed that en bloc tumor resection resulted in lower recurrence rates compared to piecemeal resection [16]. Subsequent work by Patel et al. further demonstrated that en bloc resection of intracranial metastases does not increase postoperative complications, even for large tumors or those located in eloquent areas [17]. Additionally, en bloc tumor removal reduces the risk of cerebrospinal fluid dissemination of tumor cells. To accurately define the extent of resection, auxiliary techniques such as intraoperative MRI and fluorescein-guided surgery are often employed. Evidence confirms that, compared with white light resection, fluorescence guidance significantly improves resection quality, increases the extent of resection, and prolongs overall survival (OS) [31–33]. There is also a study that shows tht patients resected using fluoresceine demonstrated significantly lower postoperative residual tumor and lower risk of local tumor recurrence, but the use of fluorescence did not influence the OS [34]. For leptomeningeal metastases, however, surgical resection holds limited therapeutic value; instead, procedures like ventriculoperitoneal shunting or Ommaya reservoir placement may be performed to alleviate symptoms, supplemented by radiotherapy or chemotherapy to delay disease progression. Jung’s retrospective analysis of 71 leptomeningeal metastasis cases found no significant survival difference between surgically and non-surgically treated groups [35], suggesting that surgery in these patients should serve primarily as palliative measure for symptom relief rather than definitive treatment. Perioperative patient management is equally critical. Patients commonly receive a range of treatments, including agents for dehydration and intracranial pressure reduction, acid suppression, fluid replacement, antiepileptics, and corticosteroids. This management is highly individualized, with treatment regimens dynamically adjusted based on a comprehensive assessment of the patient’s evolving symptoms, signs, and laboratory and imaging findings. Among these interventions, the use of corticosteroids is often inadequately emphasized. Research indicates that steroid overuse is not only associated with complications such as gastrointestinal mucosal injury and immune dysfunction but may also lead to reduced overall survival. One study demonstrated a median overall survival of 19.1 months in patients who received a cumulative dexamethasone dose of less than 122 mg, compared to 12.0 months in those receiving 122 mg or more (P = 0.002). Furthermore, multivariate analysis confirmed that a higher cumulative dexamethasone dose was independently associated with shorter survival (P < 0.001) [36]. Discussion on different types of tumors Brain metastases from lung cancer Non-small cell lung cancer (NSCLC), the most common lung cancer subtype, frequently metastasizes to the brain [18]. Its neurotropic nature [37] and the poor CNS penetration of many chemotherapy drugs contribute to historically poor outcomes, with a natural survival of 1–3 months and treated median survival of 3–6 months [38]. What’s more, TTF-1 negativity in synchronous M1b/M1c wildtype lung adenocarcinoma brain metastases predicts worse survival with increased risk of intracranial progression [39]. The historical 1-year and 2-year survival rates were only 14% and 7.6%, respectively [40]. Nevertheless, certain subgroups of patients with lung cancer brain metastases may experience a more favorable prognosis. A study of 2,186 NSCLC patients demonstrated that those diagnosed with adenocarcinoma and possessing Lung-molGPA scores between 3.5 and 4.0 achieved a median survival of nearly 4 years [41]. Current management includes surgery, radiotherapy, and systemic therapy. Complete resection of 2–3 metastases achieves local control comparable to single lesions [18]. Since systemic progression causes half of WBRT-treated NSCLC deaths [42], postoperative systemic control is vital. Our aggressive approach of en bloc resection, when feasible, yielded superior 1-year and 2-year survival rates of 48.3% and 33.2%, respectively. Figures 20, 21 and 22 confirm that complete resection of all lesions provides better long-term outcomes than debulking only symptomatic masses. Preoperative and postoperative imaging in lung cancer patients with multiple brain metastases. Preoperative head MRI demonstrated multifocal lesions in the frontal, parietal, and temporal lobes. Follow-up MRI at 6 months postoperatively showed favorable lesion control after surgical treatment Preoperative versus postoperative comparison in lung cancer patients with brain metastases. Preoperative MRI revealed a giant cystic-solid mass in the left temporal lobe, while postoperative MRI demonstrated complete tumor resection with resolution of brain shift Comparative imaging of en bloc resection for brain metastases in lung cancer patients: Preoperative versus 2-year postoperative follow-up. Preoperative MRI demonstrated a cystic-solid mass in the left parietal lobe with significant peritumoral edema. Following complete tumor resection, 2-year postoperative MRI surveillance showed no evidence of recurrence and marked reduction of perilesional edema Brain metastases from breast cancer Brain metastases develop in 9–17% of metastatic breast cancer patients, with symptomatic cases affecting 10–15% of all patients [43]; autopsy incidence reaches 30% [44]. High-risk factors include young age, high grade, large primary tumor, ER-negativity, and HER2-positivity. Subtype dictates prognosis upon brain metastasis: HER2-positive median OS is 41 months, triple-negative is 6 months, and hormone receptor-positive is 15–17 months [19]. Our cohort achieved a median survival of 17 months, demonstrating significant improvement in survival and symptoms. Figure 23 shows a representative case where complete resection of multiple metastases led to marked clinical improvement. Comparative imaging of multifocal breast cancer brain metastases before and after surgical resection. Preoperative MRI demonstrated three metastatic lesions (arrows) in the cerebellar hemisphere with significant parenchymal edema. Postoperative MRI confirmed complete resection of all three cerebellar tumors (arrows) Brain metastases from Gastrointestinal (GI) Brain metastases from gastrointestinal tumors are rare (1–2%) but portend a poor prognosis [45]. The risk rises following pulmonary metastasis, especially in left-sided colon cancer [46, 47]. While brain involvement reduces median survival in colorectal cancer from 9 to 3 months [48], our cohort achieved 10 months, consistent with literature. Given the short survival, early diagnosis is crucial. Figure 24 shows a case where complete resection of a solitary metastasis resolved symptoms and, with systemic therapy, maintained 4-year disease control. Comparative imaging before and after surgical resection for brain metastases from colorectal cancer. Preoperative MRI revealed a solid mass in the right frontal lobe with marked peritumoral edema, significant leftward midline shift, and subfalcine herniation. Postoperative MRI at 3 days confirmed complete tumor resection Brain metastases from malignant melanoma Melanoma is highly neurotropic, with approximately 37% of patients developing brain metastases—the third most common source after lung and breast cancers. Despite comprehensive treatment, median overall survival remains only about 3 months [49, 50]. There were 5 significant prognostic factors for survival: age, KPS, extracranial metastases, number of brain metastases, and BRAF status [51]. Surgical resection is the primary treatment for oligometastases, while targeted and immunotherapies represent emerging options. Our cohort achieved a median postoperative survival of 8 months, surpassing most reported outcomes and supporting the role of surgery in survival prolongation and symptom improvement. Figure 25 illustrates a case where complete resection resulted in well-controlled disease on 10-month follow-up MRI. Pre- versus post-operative MRI after en bloc metastasectomy for melanoma. Preoperative head MRI identified the tumor in the left frontal lobe. Following complete surgical resection, 10-month postoperative MRI confirmed gross total resection Study limitations Our study has several limitations. Its retrospective, single-center nature introduces potential selection bias. The lack of a control group treated non-surgically necessitates comparisons with historical data. Some subgroups were small, limiting the statistical power of certain analyses, particularly the multivariable models for gastrointestinal and melanoma cohorts. Furthermore, we lacked complete data on molecular markers (e.g., EGFR, HER2) for all patients, which are increasingly recognized as critical prognosticators. We recommend future prospective, multi-center studies to improve generalizability. Conclusion In conclusion, our study solidifies the role of maximal safe surgical resection as a cornerstone in the management of brain metastases, capable of delivering significant survival extension and neurological preservation. Patient selection, guided foremost by KPS, remains paramount. The future of neuro-oncological care lies in the sophisticated personalization and sequencing of multimodal therapy. The potential of combined preoperative and postoperative radiotherapy represents a promising avenue for local control intensification, while the integration of targeted and systemic therapies is non-negotiable for maximizing survival. Ultimately, treatment must be tailored through a multidisciplinary lens, with surgery serving as a powerful intervention within a holistic, patient-centric strategy. Author contributions Peng Sun is responsible for the organization of the thesis materials and the writing of the manuscript. Chunyu Gu, Haoran Wang, Yanzhen Zhang, Jianping Cao, Junpeng Ji, Mingshan Zhang are responsible for the verification and organization of the article data. Zhang Mingshan is responsible for the review of the manuscript and the data. Funding No funding. Data availability All data generated or analysed during this study are included in this published article. Declarations Ethics approval and consent to participate The study protocol was approved by the Ethics Committee of Sanbo Brain Hospital, Capital Medical University. All research procedures were strictly conducted in accordance with the guidelines established by this ethics committee and conformed to the ethical principles of the Declaration of Helsinki. We confirm that all research operations involving human participants, including but not limited to the collection and analysis of patient clinical data and imaging acquisition, were performed after obtaining written informed consent signed by the participants or their legal guardians. Throughout the research process, we remain committed to protecting participants’ privacy rights and personal data security. The participant has consented to the submission of the case report to the journal. Patients signed informed consent regarding publishing their data and photographs. Consent for publication All the listed authous have carefully reviewed and approved this manuscript. We all agree to submit to Discover Oncology for publication and are aware of and accept the journal’s publication policies. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement All data generated or analysed during this study are included in this published article.