Skip to main content
Download PDF

Integrated analysis of molecular and clinical features associated with overall survival in melanoma patients with brain metastasis

Acta Neuropathologica Communications volume 13, Article number: 75 (2025) Cite this article

Abstract

Melanoma brain metastases (MBMs) are diagnosed in up to 60% of metastatic melanoma patients. Previous studies have identified clinical factors that correlate with overall survival (OS) after MBM diagnosis. However, molecular and immune features associated with OS are poorly understood. An improved understanding of the molecular and immune correlates of OS could provide insights into MBM patient outcomes and guide therapeutic development. Thus, we analyzed clinical features and outcomes of 74 melanoma patients who underwent surgical resection (via craniotomy) between 1991 and 2015 at our institution with RNA-seq data generated from their MBMs. The median post-operative OS was 8.6 months (range 0.6–146.9). On univariate analysis (UVA), the expression of multiple immune gene signatures was associated with improved OS, including IFN-γ Index, T cell-inflamed and the Expanded Immune Genes. The gene expression signatures of several immune cell types (i.e., T cells, CD8 T cells, cytotoxic lymphocytes, NK cells, monocytes) positively correlated with OS, whereas higher neutrophil gene expression correlated with shorter OS. UVA of clinical features identified low Karnofsky performance score (KPS), elevated serum lactate dehydrogenase (LDH), presence of extracranial metastases (ECMs), and uncontrolled (versus controlled) ECMs as clinical predictors of shorter survival. Multivariate analyses (MVA) were performed with significant clinical factors and all immune features without any redundant highly correlated variables in the model. After backward selection, multivariable coxPH model identified low KPS, low T cell signature, and low monocytic lineage signature as independent predictors of shorter survival. Finally, comparative analysis of MBMs from patients with MBMs only showed that these tumors were characterized by decreased oxidative phosphorylation (OXPHOS) and increased immune infiltration signature versus MBMs from patients with concurrent ECMs. Together these results support the clinical significance of specific immune features of MBMs and suggest their potential use as prognostic biomarkers.

Introduction

Melanoma is the most aggressive of the common forms of skin cancer. Melanoma accounts for 70% of skin-cancer-related deaths, even though it comprises only 5% of skin cancer cases [1, 2]. Brain metastases, one of the most lethal sites of distant metastasis, are diagnosed in up to 60% of Stage IV melanoma patients and up to 80% of patients at autopsy [3]. Historically, and particularly prior to the development of effective systemic treatments for this disease, patients with melanoma brain metastasis (MBM) had a median overall survival (OS) of approximately four months [4, 5]. Fortunately, outcomes for MBM patients have improved over the last decade, coinciding with the development of more effective focal radiation techniques and new immune and targeted therapies [6]. However, ultimately the majority of MBM patients still die of their disease.

We and others have recently shown by broad molecular profiling (i.e., RNA-sequencing) of patient-matched samples from individual patients that MBMs consistently harbor unique signaling, immune, and/or metabolic features compared to extracranial metastases (ECMs) [2]. Unique features identified in brain metastases from melanoma include the upregulation of oncogenic signaling pathways (i.e., PI3K-AKT pathway) [7,8,9], enrichment of metabolic pathways (i.e., OXPHOS, glucose-derived serine biosynthesis, fatty acid synthesis) [2, 10,11,12,13,14,15] and decreased infiltration by several immune cell populations (i.e., CD3 + cells, CD8 + cells, monocytic lineage cells, myeloid dendritic cells) [2] compared to ECMs and primary tumors. In parallel, there have been several investigations of clinical features that correlate with OS in MBM patients [5, 6, 16,17,18]. These studies have identified several clinical features that consistently associate with OS including Karnofsky performance status (KPS), age, post-operative radiation therapy, and presence of extracranial disease. However, there have been very few efforts to integrate the molecular characterization of MBMs with clinical features and patient outcomes. Such efforts could improve risk modeling in these patients, which can inform patient management and/or improve clinical trial design. This approach could also provide new insights into the basis of clinical features that have frequently been associated with OS in MBM patients. Together these approaches may translate into new approaches for these highly aggressive tumors.

Thus, here we present the integrated analysis of global RNA-sequencing data, clinical features, and OS in a cohort of 74 MBM patients. In addition to identifying molecular and clinical features associated with OS in this integrated analysis, we explored the differences between MBMs from patients with and without extracranial disease.

Methods

Patient cohort and clinical data

The cohort included 74 MBMs patients who underwent a craniotomy for surgical resection of a melanoma brain metastasis between July 1991 and October 2015 at our institution whose tumors were analyzed by RNA-sequencing, as previously described [2]. Following a literature review [16,17,18,19,20,21,22], clinical features recurringly associated with OS in MBM patients treated with surgical resection were collected under an Institutional Review Board approved study. Clinical features collected included age, gender, pre-operative KPS, post-operative radiation, number of brain metastases, presence of extracranial disease, status of extracranial disease, surgical extent of resection, and serum lactate dehydrogenase (LDH) (elevated versus not, within 30 days of surgery). Status of extracranial disease was determined by body PET/CT scan or CT scan performed within 6 weeks of surgery or last follow-up date. Uncontrolled systemic disease was defined as new or progressive extracranial tumor burden. Patient with newly diagnosed metastatic disease were also diagnosed as uncontrolled. Surgical extent of resection was categorized as gross total resection (GTR) or subtotal resection (STR) based on the post-operative contrast MRI radiology report.

RNA-seq data processing

Briefly, and as described previously [2], RNA was extracted from regions containing 70% or more viable tumor cells on FFPE tissue blocks. RNA was sequenced using HiSeq 4000 76-bp paired-end sequencing. Pre-processed gene by raw count matrix was available from our previous publication [2, 15]. The count data for all the samples were converted into counts-per-million (CPM) data using edgeR package in R and low expressed genes (CPM < 0.5) were filtered out. The trimmed mean of M values (TMM) normalization and Voom transformation steps were performed to eliminate library size bias and to normalize transcript counts. This normalized data was used for Gene Set Enrichment Analysis (GSEA) [23] and different types of gene signature analysis explained below. Single sample GSEA (ssGSEA) for various gene sets were performed on TMM-normalized, Voom-transformed logCPM expression matrix using GenePattern module ssGSEAProjection. As an example, OXPHOS (OP)-Index, an average of ssGSEA scores of 8 OXPHOS-related pathways, was calculated as previously described [2, 15].

For patients with RNA-seq data from MBMs from more than one craniotomy, the data from the earliest resected MBM was included in data modeling. For patients with data from more than one MBM at the selected craniotomy, random sampling of the two MBM was chosen using R programming set.seed(1) function.

Immune-gene signature and immune-cell deconvolution analysis

The gene expression data was used to generate immune-based gene signatures; ImmuneScore using ESTIMATE [24] R package, IFN-γ Index, and expanded immune gene signature and T cell-inflamed signature [25] as described previously. ImmuneScore was generated from 141 immune cell related gene expression values and predicts the proportion of immune cells in the sample. 6-gene based IFN-γ Index, 18-gene based expanded immune gene signature, 18-gene based T cell–inflamed gene expression profile (GEP) correlates with improved response to anti-PD-1 immunotherapy and contains IFN-γ responsive genes related to antigen presentation, chemokine expression, cytotoxic activity, and adaptive immune resistance. MCP-counter [26], a bioinformatic deconvolution tool, was used in R to predict ten different (eight are immune-related) cell populations present in each sample from its gene expression data. For molecular and Immune features, the above-mentioned gene signatures and cell type scores were used.

Univariate and multivariate analysis

OS was computed from the date of craniotomy to the date of last known vital status. Patients alive at last follow up were censored. Cox Proportional Hazard (coxPH) modeling was used for univariate and multivariate analysis of associations with OS. For univariate analysis, each clinical feature was considered separately, and each molecular feature was considered separately to test for significant association with OS. Before performing the univariate analysis, all clinical, molecular, and immune features which were continuous variables were normalized and scaled using the scale function in R. Clinical and immune features that were p < 0.1 by univariate analysis were combined, without any redundant highly correlated variables in the original coxPH multivariate model. Features with correlation coefficient r (Pearson product moment) or rho (Spearman rank) value >|0.7| was considered as highly correlated variables. Both the original and final coxPH model were tested to confirm that the proportional hazards assumption was true (global p < 0.05). The final multivariate model was selected when all the variables in the coxPH model were significant (p < 0.05) predictors of OS after backward feature selection technique [27]. Backward feature selection was performed by removing one feature at a time. The feature with the highest p-value after each multivariate analysis was removed. This step was performed iteratively until all features in the final coxPH model had significant p-value < 0.05.

Hazard Ratio (HR) and 95% Confidence Interval (CI) were calculated for each feature. HR above 1 was considered to be negatively correlated with OS and HR below 1 was considered to be positively correlated with OS. All the analysis were performed using R v4.2.3 in R studio software.

Results

Differential gene expression and pathways between mbm patients with long versus short term survival

We performed initial exploratory analyses to identify gene expression pathways differentially expressed between patients with long-term vs short-term OS from MBM surgical resection (Fig. 1A). For this analysis, we compared patients in the top tercile of OS duration (Long-Term Survivors (LTS), n = 26, median OS 22.6 months) versus the bottom tercile (Short-Term Survivors (STS), n = 25, median OS 2.6 months). The difference in OS between the LTS and STS patients was highly significant (HR = 0.03, 95% confidence interval (CI) 0.01–0.07; p < 0.0001) (Fig. 1B).

Fig. 1
figure 1

Differentially expressed pathways in MBM long term survivors (LTS) compared to short-term survivors (STS). A. Schematic of the analysis of MBM LTS versus STS gene expression patterns. For patients with RNA expression data from more than one surgically resected MBM, the earliest resected sample was used for analysis. B. Kaplan–Meier OS curves for LTS (solid line) and STS (dashed line) MBM patients with Hazard ratio (HR) and 95% Confidence interval. Median OS is 22.6 months for the LTS group and 2.6 months for the STS group. C. Analysis of differentially expressed Hallmark pathways demonstrates that inflammatory- and interferon-related gene sets were upregulated (in black), while MYC targets and OXPHOS-related pathways were downregulated (in grey), in LTS compared to STS MBMs. X-axis is Negative Log FDR (False Discovery Rate). D. Boxplot of OXPHOS Index for LTS (circles) vs STS (triangles) MBMs

Full size image

Differential gene expression analysis of the RNA-seq for MBMs from the two groups of patients was performed. Twelve genes were downregulated, and 187 genes were upregulated significantly (FDR < 0.05) in LTS compared to STS (Supplementary Table 1, Excel Sheet No. 1). Several immune related genes (i.e., Interleukin receptors 12, 21; MHC class II molecules) were enriched in the LTS patients. Gene Set Enrichment Analysis (GSEA) [23] using gene sets such as gene ontology pathways, Hallmark pathways, C2_curated gene sets, and C7_immunologic_signatures gene sets (Supplementary Table 1, Excel Sheet No. 2–5) were also performed. Twelve out of 50 Hallmark pathways were differently expressed between the two groups of MBMs (Fig. 1C). Inflammatory- and interferon-related gene sets were upregulated in MBMs of the LTS patients, while MYC targets and OXPHOS-related pathways (Fig. 1D) were downregulated in the LTS MBMs compared to STS MBMs. Other GSEA gene sets results corroborated immune enrichment in the LTS MBMs compared to STS MBMs (Supplementary Table 1, Excel Sheet No. 2–5).

As the global pathway analyses implicated immune enrichment in MBMs from LTS patients, we performed additional analyses to compare the immune infiltrates and gene signatures of the two groups. The ImmuneScore, a signature based on the expression of 141 immune cell related genes, was enriched in the MBMs from LTS patients (p < 0.001; Fig. 2A). LTS MBMs also exhibited enrichment of signatures of Expanded Immune Signature, IFN-γ Index, and T cell-inflamed Signature gene sets, which have been associated with responsiveness to immune checkpoint inhibitors (p < 0.001; Fig. 2B). Estimation of the prevalence of immune cell populations by MCP-counter [26] showed significant enrichment of signatures of T cells (p = 1.6e–04), CD8 T cells (p = 0.0250), cytotoxic lymphocytes (p = 7.8e–03), NK cells (p = 5.1e–03), and Monocytic lineage (p = 8.6e–04) cells in MBMs from the LTS patients (Fig. 2C). No significant differences were detected for B cells, Myeloid dendritic cells, or Neutrophils between the two groups. Previously performed IHC [2] confirmed increased positivity for CD3( +) (p = 0.0015) and CD8( +) (p = 0.039) cells in the LTS cohort, and no significant difference in PAX5( +) (B cell marker; p = 0.31) (Supplementary Fig. 1).

Fig. 2
figure 2

Differential immune gene signatures and cell types between LTS and STS MBMs. A. Boxplot of ImmuneScore for LTS (circles) and STS (triangles) MBMs. B. Boxplot of expanded immune gene signature, IFN-γ Index, T cell-inflamed gene signature for LTS (circles) vs STS (triangles) MBMs. C. MCP-counter estimates of the prevalence of indicated cell populations in LTS (circles) and STS (triangles) MBMs. p < 0.05 was considered significant by Wilcoxon Rank Sum test

Full size image

Features associated with overall survival in the full cohort

After this exploratory comparison of LTS versus STS MBM patients, we performed univariate and multivariate analyses of factors associated with OS using all patients (n = 74) in the cohort (Fig. 3). When one sample per patient was considered, the median OS from the first craniotomy was 8.6 months (95% CI, 6.5–13.6 months) (Supplementary Fig. 2). The exploratory analysis showed that most differences between LTS vs STS were immune-related. This helped in choosing immune features such as immune gene signatures and predicted composition of immune cell types for survival analysis. Clinical features of the cohort are shown in Table 1. Overall, 58.1% were male, and the median age was 54. The median KPS at the time of surgery was 90 (n = 60 patients with KPS documentation).

Fig. 3
figure 3

Analysis schema to identify prognostic clinical, molecular and immune prognostic factors in MBM patients. First, univariate analysis of indicated clinical factors, molecular features, and immune features were performed separately. Features with p < 0.1 on univariate analysis were utilized in multivariate analysis to identify independent predictors of OS in MBM patients

Full size image
Table 1 Univariate analysis of associations of clinical factors with overall survival. P-value was calculated using log rank sum test. Hazard Ratio (HR) > 1 indicates negative correlation with OS.
Full size table

Univariate analysis of clinical features identified several factors significantly associated with post-operative OS, including pre-operative KPS (Continuous score of 20–100, increases by 10; Hazard Ratio (HR) 0.98, 95% Confidence Interval (CI) 0.96 – 0.99, p = 0.006), KPS > 70 (No vs Yes: HR 0.41, 95% CI 0.19–0.89, p = 0.025), presence of extracranial disease (HR 2.22, 95% CI 1.22–4.03, p = 0.008), and uncontrolled extracranial disease status (HR 2.13, 95% CI 1.25–3.63, p = 0.006), serum LDH (HR 0.53, 95% CI 0.28–0.98, p = 0.042) (Table 1). Age, gender, post-operative radiation, extent of resection, presence of more than one MBM were not significantly associated with post-operative OS (Table 1).

Univariate analysis of the RNA-seq data for all MBMs patients showed that multiple immune signature scores (normalized and scaled) were associated with improved post-operative OS, including ImmuneScore (HR 0.69, 95% CI 0.51–0.93, p = 0.014), IFN-γ index (HR 0.54, 95% CI 0.41–0.72, p < 0.001), Expanded Immune Gene Signature (HR 0.59, 95% CI 0.45–0.78, p < 0.001), and T cell-inflamed Signature (HR 0.57, 95% CI 0.44–0.76, p < 0.001) (Table 2). MCP-counter analysis identified that signatures of increased T cells (HR 0.65, 95% CI 0.52–0.83, p < 0.001), CD8( +) T cells (HR 0.78, 95% CI 0.62–0.98, p = 0.035), cytotoxic lymphocytes (HR 0.72, 95% CI 0.55–0.94, p = 0.017), NK cells (HR 0.71, 95% CI 0.54–0.95, p = 0.019), and monocytic lineage (HR 0.69, 95% CI 0.52–0.92, p = 0.012) cells were associated with improved OS, while increased neutrophils (HR 1.33, 95% CI 0.99–1.79, p = 0.057) were associated with shorter OS (Table 2). B lineage, Myeloid dendritic cells, Endothelial cells, and Fibroblasts were not significantly associated with OS (Table 2).

Table 2 Univariate analysis of associations of normalized molecular and immune features with overall survival. P-value was calculated using log rank sum test. Hazard Ratio (HR)> 1 indicates negative correlation with OS.
Full size table

Multivariable Cox proportional hazard analysis was performed including all features (clinical and molecular) that were significant (p < 0.1) on univariate analysis. Pairwise comparison of correlation coefficient of all clinical and molecular features identified IFN-γ index, Expanded Immune Gene Signature, and T cell-inflamed Signature to be highly correlated variables, and only one of them (IFN-γ index) was decided to be included in the original multivariate coxPH model. Since IFN-γ index was highly (correlation coefficient > 0.7) correlated with ImmuneScore and T cells signature, IFN-γ Index was not also included, but ImmuneScore and T cells signature were included in the original model since these two variables were not highly correlated among themselves. The original coxPH model, before backward feature selection algorithm was applied, included pre-operative KPS, extracranial disease status, serum LDH, normalized ImmuneScore, and normalized MCP-Counter Predicted Cell types: T cells, CD8 T cells, cytotoxic lymphocytes, NK cells, monocytic lineage, and neutrophils (Table 3). Multivariate analysis of the original coxPH model identified pre-operative KPS (HR 0.96, 95% CI 0.94–0.99, p = 0.002), T cells (HR 0.50, 95% CI 0.32–0.79, p = 0.003), monocytic lineage (HR 0.32, 95% CI 0.11–0.91, p = 0.032), and neutrophils (HR 1.82, 95% CI 1.12–2.96, p = 0.016) as significant independent predictors of post-operative OS in MBM patients (Table 3). Backward feature selection algorithm was applied to the all the variables in the original model as described in the ‘Methods’ and the final coxPH model resulted in three significant independent of OS in MBMs: pre-operative KPS (HR 0.95, 95% CI 0.94–0.97, p < 0.001), normalized T cells (HR 0.59, 95% CI 0.43–0.80, p < 0.001), and normalized monocytic lineage (HR 0.55, 95% CI 0.37–0.81, p = 0.003) (Table 4).

Table 3 Multivariate analysis of association of clinical factors and normalized immune features with post-operative overall survival, before performing backward feature technique for model selection. Factors with p< 0.1 on univariate analysis were included in the original multivariate coxPH model. P-value was calculated using log rank sum test. Features with p < 0.05 were considered as independent predictors of OS. Hazard Ratio (HR) > 1 indicates negative correlation with OS.
Full size table
Table 4 Multivariate analysis of association of clinical factors and normalized immune features with post-operative overall survival, after performing backward feature technique for model selection. P-value was calculated using log rank sum test. Features with p < 0.05 were considered as independent predictors of OS. Hazard Ratio (HR) > 1 indicates negative correlation with OS.
Full size table

Molecular differences between mbm patients with versus without extracranial disease

As reported previously in other studies [4,5,6], we also observed improved post-operative OS in MBM patients without concurrent extracranial metastases (ECMs) (i.e. with brain metastases only) (Table 1 and Supplemental Fig. 3). While improved outcomes in such patients might be due to the need only for local therapies (i.e., surgery, radiation) to render them free of disease, it is currently unknown if these tumors also have molecular features that could explain their improved outcomes. Thus, we compared RNA-seq data of MBMs from patients with brain metastases only (n = 29) versus those with concurrent ECMs (n = 56) (Fig. 4A). Differential gene expression analysis of the RNA-seq for the two groups of tumors identified three significantly downregulated genes and 187 genes significantly upregulated genes (FDR < 0.05) (Supplementary Table 2, Excel Sheet No. 1) in MBMs from patients without concurrent ECMs. Gene Set Enrichment Analysis (GSEA) [23] using gene sets such as gene ontology pathways (67 regulated), Hallmark pathways (7 upregulated and 2 downregulated), C2_curated gene sets (264 upregulated and 23 downregulated), and C7_immunologic_signatures gene sets (135 upregulated and one downregulated) (Supplementary Table 2, Excel Sheet No. 2–5) were also performed. GSEA analysis results indicated immune enrichment in the patients with MBMs only compared to MBM patients with concurrent ECMs (Supplementary Table 2, Excel Sheet No. 2–5).

Fig. 4
figure 4

Gene signatures and cell types that are different between MBM-only versus MBM-with extracranial metastases (ECMs). A. Schema of the analysis. B. Boxplot of OXPHOS Index for MBM-only (circles) and MBM-with_ECM (triangles) tumors. p < 0.05 was considered significant by Wilcoxon Rank Sum test. C. Boxplot of expression of indicated Immune-activation gene signature scores for MBM-only (circles) and MBM-with-ECM (triangles) tumors. D. Boxplot of ImmuneScores for MBM-only (circles) and MBM-with_ECM (triangles) tumors. E. MCP-counter analysis of prevalence of indicated cell types in MBM-only (circles) and MBM-with-ECM (triangles) tumors. p < 0.05 was considered significant by Wilcoxon Rank Sum test

Full size image

GSEA analysis also demonstrated that MBMs from patients without concurrent ECMs had lower expression of OXPHOS genes, which was also detected by determination of the OXPHOS_index (p = 0.011, Fig. 4B); as well as increased expression of the favorable immune gene signatures and ImmuneScore (p < 0.05, Fig. 4C, D). Finally, MCP-counter analysis showed that MBM-only tumors had increased T cells (p = 0.03), cytotoxic lymphocytes (p = 0.03), NK cells (p = 0.002), myeloid dendritic cells (p = 0.001), and endothelial cells (p = 0.03) (Fig. 4E). No significant differences were detected for CD8 T cells, B cells, monocytic lineage, neutrophils or fibroblasts between the two groups (p > 0.05, Fig. 4E).

Discussion

Recent studies by several groups have provided novel insights into the unique features and heterogeneity of brain metastases from melanoma and other cancers [6, 11, 28,29,30]. Interestingly, many of the features that are enriched in brain metastases appear to be shared across different tumor types [28, 29]. Efforts are now ongoing to try to leverage such information into new therapies for patients with brain metastases [6, 31]. There is also a rationale to evaluate if/how these features correlate with clinical outcomes. Thus, we have conducted one of the first integrated, multivariable analyses to combining global molecular profiling by RNA-sequencing with clinical prognostic factors, to determine the impact of OS in melanoma patients with brain metastases.

Many studies [4,5,6] that have described clinical features that are associated with OS in MBM patients. However, very few have performed integrated analyses that include the molecular features of these tumors. Further, most such studies have used relatively focused profiling approaches (i.e., individual gene mutations, immunohistochemistry) [7, 32,33,34,35] to correlate with patients’ survival. Studies that have performed multivariate analysis [7, 35] of clinical features along with molecular features are limited. Even in studies that have analyzed molecular characteristics, the data has historically been limited in scope e.g. DNA-based single gene [5, 6, 32, 35] or a set of gene mutations [8, 9, 36], histopathological HE stains [7] and/or immunohistochemistry (IHC) staining [33, 34, 37] of a limited number of cell surface markers. To our knowledge, ours is the first study to integrate RNA-seq data with clinical characteristics in a multivariate analysis to identify features associated with OS in MBM patients. Our current study shows that even in the context of robust clinical prognostic factors, the immune features of MBMs are strongly associated with OS.

While the brain has often been considered an immune-privileged site, our analyses consistently demonstrate a positive correlation between the gene expression signatures of multiple immune cell populations with post-operative survival among this cohort of MBM patients. This result is consistent with a previous study of 29 melanoma patients that included IHC-based assessment of immune cell populations in MBMs. In that study, a multivariable analysis that included clinical features demonstrated that “immune infiltrate” was a significant predictor of improved post-operative survival in MBMs patients [37]. The use of RNA-seq in our cohort allows for broader interrogation of immune cell populations and signatures, and the availability of a larger number of patients provides additional value and confidence in the results. Notably, while pre-operative KPS, T-cells, and monocytic lineage were associated with improved OS, neutrophils were associated with worse post-operative OS in our analysis, which was not analyzed in the previous IHC study [37]. However, the finding presented in our study is consistent with studies that have associated neutrophils with tumor progression and therapeutic resistance in melanoma patients, albeit in the setting of non-CNS tumors [38]. In our final multivariate coxPH model after backward feature selection, we identified pre-operative KPS, T-cell gene signature, and monocytic lineage gene signature as independent predictors of post-operative OS in MBM patients. These results support the inclusion of patients with MBMs in efforts to alter the infiltration and activation of the above significant immune cell types to improve outcomes in this disease.

As expected, and consistent with many previous studies, we also observed that melanoma patients with brain metastases only had better OS than those patients with concurrent ECMs [39, 40]. While there are likely several clinical and treatment factors for this association, we observed several molecular differences in these tumors compared to MBMs from patients with concurrent extracranial disease. Tumors from MBM-only patients featured gene expression consistent with increased infiltration of several immune populations and increased expression of favorable immune gene signatures, which could confer improved responsiveness to host immune responses and immune therapies [25, 31, 41]. Tumors from MBM-only patients also featured decreased expression of OXPHOS genes. This may again reflect and/or drive a more favorable tumor biology, as OXPHOS has been associated with resistance to multiple therapies, including both immune and targeted therapies for melanoma [42,43,44]. Notably, previous work has demonstrated that decreased tumor OXPHOS can results in improved activation of infiltrating immune cells [15, 45], and thus is consistent with the association and difference observed here. It will be interesting to determine if similar findings are observed in other cohorts of MBM patients, or in patients with brain metastases from other cancer types.

Since MBM-only patients have better post-operative OS compared to MBM patients with ECMs and better OS is associated with higher levels of immune gene signatures and immune cell infiltrations [2, 6], it is conceivable that MBMs-only patients may have a more favorable immune microenvironment in their brain tumors. Lower level of OXPHOS might also be contributing to increased immune activation within MBMs-only tumors. Notably, our ongoing work (manuscript in preparation) indicates that OXPHOS is a marker of aggressive tumors and correlates with increased risk of distant metastasis overall (i.e., not only to the CNS).

We acknowledge that our study has limitations. First, our analysis was limited to MBM patients who underwent a surgical resection. Thus, the results may not reflect all MBM patients, particularly those with smaller tumors that are usually treated non-invasively (i.e., with radiation or systemic therapy) or those with low functional status and unsuitable for an invasive procedure Second, based on the craniotomy dates of the MBMs included in this analysis, very few patients in this cohort received treatment with contemporary immunotherapy (such as immune checkpoint blockade, n = 13 patients, 15 MBMs) or targeted therapies (BRAFi/MEKi, n = 7 patients, 3 MBMs). Notably, studies [46] that include patients that have received such treatments generally only characterize MBMs that failed to respond to those therapies, as it is very rare to resect a tumor that is responding to treatment except in the setting of window-of-opportunity clinical trials. Thus, it is unclear if prognostic molecular features in this cohort of patients that had surgery are also predictive of outcomes in MBM patients receiving other therapies. This limitation could potentially be addressed through the development of technologies that allow for non-invasive assessment of immune infiltrates or gene signatures, such as imaging tracers and/or radiomics. Importantly, our study suggests that such studies may need to be designed to examine specific immune populations or pathways. We also acknowledge that the patients in our study are heterogenous in the treatments received before and/or after surgery. The size of our cohort does not provide sufficient statistical power to appropriately assess for differences in outcomes based on the treatments received. The associations observed here also merit testing in other cohorts of MBM patients and in patients with brain metastases from other cancer types. Finally, while bulk RNA-sequencing gives us global molecular profiling of gene expression, additional insights may be facilitated by single-cell approaches, it may not be ideal method identify specific immune cell population changes or metabolic changes.

Conclusions

In summary, this study is one of the first to integrate the results of whole genome transcriptional profiling with relevant clinical factors in assessing prognostic markers in patients with MBMs- and one of the largest such studies in any cancer type. Our analyses demonstrate the prognostic significance of immune infiltrates, including specific immune cell populations, in MBMs even when accounting for robust/traditional clinical prognostic factors. Unexpectedly, we also identified specific immune and molecular features of MBMs from patients without concurrent extracranial disease. Together the findings support the importance of, and indicate the rationale for continued investigations into, the key regulators of the immune response in MBMs.

Availability of data and materials

Datasets used in this manuscript comes from our previous study “Fischer GM, Jalali A, Kircher DA, et al. Molecular Profiling Reveals Unique Immune and Metabolic Features of Melanoma Brain Metastases. Cancer Discov. 2019; 9(5):628–645.”

Abbreviations

MBMs:

Melanoma Brain Metastases

ECMs:

Extracranial Metastases

UVA:

Univariate Analysis

MVA:

Multivariate Analyses

KPS:

Karnofsky Performance Score

LDH:

Lactate Dehydrogenase

OXPHOS:

Oxidative Phosphorylation

OS:

Overall Survival

GTR:

Gross Total Resection

STR:

Subtotal Resection

GSEA:

Gene Set Enrichment Analysis

coxPH:

Cox Proportional Hazard

HR:

Hazard Ratio

CI:

Confidence Interval

LTS:

Long-Term Survivors

STS:

Short-Term Survivors

IHC:

Immunohistochemistry

FDR:

False Discovery Rate

References

  1. Tripp MK, Watson M, Balk SJ, Swetter SM, Gershenwald JE (2016) State of the science on prevention and screening to reduce melanoma incidence and mortality: the time is now. CA Cancer J Clin 66(6):460–480

    Article  PubMed  PubMed Central  Google Scholar 

  2. Fischer GM, Jalali A, Kircher DA et al (2019) Molecular profiling reveals unique immune and metabolic features of melanoma brain metastases. Cancer Discov 9(5):628–645

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  3. Karz A, Dimitrova M, Kleffman K et al (2022) Melanoma central nervous system metastases: an update to approaches, challenges, and opportunities. Pigment Cell Melanoma Res 35(6):554–572

    Article  PubMed  PubMed Central  Google Scholar 

  4. Davies MA, Liu P, McIntyre S et al (2011) Prognostic factors for survival in melanoma patients with brain metastases. Cancer 117(8):1687–1696

    Article  PubMed  Google Scholar 

  5. Raizer JJ, Hwu WJ, Panageas KS et al (2008) Brain and leptomeningeal metastases from cutaneous melanoma: survival outcomes based on clinical features. Neuro Oncol 10(2):199–207

    Article  PubMed  PubMed Central  Google Scholar 

  6. Hasanov M, Milton DR, Davies AB et al (2023) Changes in outcomes and factors associated with survival in melanoma patients with brain metastases. Neuro Oncol 25(7):1310–1320

    Article  PubMed  CAS  Google Scholar 

  7. Chen G, Chakravarti N, Aardalen K et al (2014) Molecular profiling of patient-matched brain and extracranial melanoma metastases implicates the PI3K pathway as a therapeutic target. Clin Cancer Res 20(21):5537–5546

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  8. Niessner H, Forschner A, Klumpp B et al (2013) Targeting hyperactivation of the AKT survival pathway to overcome therapy resistance of melanoma brain metastases. Cancer Med 2(1):76–85

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  9. Seifert H, Hirata E, Gore M et al (2016) Extrinsic factors can mediate resistance to BRAF inhibition in central nervous system melanoma metastases. Pigment Cell Melanoma Res 29(1):92–100

    Article  PubMed  CAS  Google Scholar 

  10. Kleffman K, Levinson G, Rose IVL et al (2022) Melanoma-secreted amyloid beta suppresses neuroinflammation and promotes brain metastasis. Cancer Discov 12(5):1314–1335

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  11. Biermann J, Melms JC, Amin AD et al (2022) Dissecting the treatment-naive ecosystem of human melanoma brain metastasis. Cell 185(14):2591–2608

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  12. Ngo B, Kim E, Osorio-Vasquez V et al (2020) Limited environmental serine and glycine confer brain metastasis sensitivity to PHGDH inhibition. Cancer Discov 10(9):1352–1373

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  13. Zou Y, Watters A, Cheng N et al (2019) Polyunsaturated fatty acids from astrocytes activate PPARgamma Signaling in cancer cells to promote brain metastasis. Cancer Discov 9(12):1720–1735

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  14. Ferraro GB, Ali A, Luengo A et al (2021) Fatty acid synthesis is required for breast cancer brain metastasis. Nat Cancer 2(4):414–428

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  15. Fischer GM, Guerrieri RA, Hu Q et al (2021) Clinical, molecular, metabolic, and immune features associated with oxidative phosphorylation in melanoma brain metastases. Neurooncol Adv. https://doiorg.publicaciones.saludcastillayleon.es/10.1093/noajnl/vdaa177

    Article  PubMed  PubMed Central  Google Scholar 

  16. Patchell RA, Tibbs PA, Walsh JW et al (1990) A randomized trial of surgery in the treatment of single metastases to the brain. N Engl J Med 322(8):494–500

    Article  PubMed  CAS  Google Scholar 

  17. Lee CH, Kim DG, Kim JW et al (2013) The role of surgical resection in the management of brain metastasis: a 17 year longitudinal study. Acta Neurochir (Wien) 155(3):389–397

    Article  PubMed  Google Scholar 

  18. Mahajan A, Ahmed S, McAleer MF et al (2017) Post-operative stereotactic radiosurgery versus observation for completely resected brain metastases: a single-centre, randomised, controlled, phase 3 trial. Lancet Oncol 18(8):1040–1048

    Article  PubMed  PubMed Central  Google Scholar 

  19. Vecht CJ, Haaxma-Reiche H, Noordijk EM et al (1993) Treatment of single brain metastasis: radiotherapy alone or combined with neurosurgery? Ann Neurol 33(6):583–590

    Article  PubMed  CAS  Google Scholar 

  20. Brown PD, Ballman KV, Cerhan JH et al (2017) Postoperative stereotactic radiosurgery compared with whole brain radiotherapy for resected metastatic brain disease (NCCTG N107C/CEC·3): a multicentre, randomised, controlled, phase 3 trial. Lancet Oncol 18(8):1049–1060

    Article  PubMed  PubMed Central  Google Scholar 

  21. Fischer GM, Carapeto FCL, Joon AY et al (2020) Molecular and immunological associations of elevated serum lactate dehydrogenase in metastatic melanoma patients: a fresh look at an old biomarker. Cancer Med 9(22):8650–8661

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  22. Gaspar L, Scott C, Rotman M et al (1997) Recursive partitioning analysis (RPA) of prognostic factors in three Radiation Therapy Oncology Group (RTOG) brain metastases trials. Int J Radiat Oncol Biol Phys 37(4):745–751

    Article  PubMed  CAS  Google Scholar 

  23. Subramanian A, Tamayo P, Mootha VK et al (2005) Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 102(43):15545–15550

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  24. Yoshihara K, Shahmoradgoli M, Martinez E et al (2013) Inferring tumour purity and stromal and immune cell admixture from expression data. Nat Commun 4:2612

    Article  PubMed  Google Scholar 

  25. Ayers M, Lunceford J, Nebozhyn M et al (2017) IFN-gamma-related mRNA profile predicts clinical response to PD-1 blockade. J Clin Invest 127(8):2930–2940

    Article  PubMed  PubMed Central  Google Scholar 

  26. Becht E, Giraldo NA, Lacroix L et al (2016) Estimating the population abundance of tissue-infiltrating immune and stromal cell populations using gene expression. Genome Biol 17(1):218

    Article  PubMed  PubMed Central  Google Scholar 

  27. Kuk D, Varadhan R (2013) Model selection in competing risks regression. Stat Med 32(18):3077–3088

    Article  PubMed  Google Scholar 

  28. Fukumura K, Malgulwar PB, Fischer GM et al (2021) Multi-omic molecular profiling reveals potentially targetable abnormalities shared across multiple histologies of brain metastasis. Acta Neuropathol 141(2):303–321

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  29. Alvarez-Prado AF, Maas RR, Soukup K et al (2023) Immunogenomic analysis of human brain metastases reveals diverse immune landscapes across genetically distinct tumors. Cell Rep Med 4(1):100900

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  30. Brastianos PK, Carter SL, Santagata S et al (2015) Genomic characterization of brain metastases reveals branched evolution and potential therapeutic targets. Cancer Discov 5(11):1164–1177

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  31. Tumeh PC, Harview CL, Yearley JH et al (2014) PD-1 blockade induces responses by inhibiting adaptive immune resistance. Nature 515(7528):568–571

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  32. Jakob JA, Bassett RL Jr, Ng CS et al (2012) NRAS mutation status is an independent prognostic factor in metastatic melanoma. Cancer 118(16):4014–4023

    Article  PubMed  CAS  Google Scholar 

  33. Berghoff AS, Fuchs E, Ricken G et al (2016) Density of tumor-infiltrating lymphocytes correlates with extent of brain edema and overall survival time in patients with brain metastases. Oncoimmunology 5(1):e1057388

    Article  PubMed  Google Scholar 

  34. Harter PN, Bernatz S, Scholz A et al (2015) Distribution and prognostic relevance of tumor-infiltrating lymphocytes (TILs) and PD-1/PD-L1 immune checkpoints in human brain metastases. Oncotarget 6(38):40836–40849

    Article  PubMed  PubMed Central  Google Scholar 

  35. Vasudevan HN, Delley C, Chen WC et al (2023) Molecular features of resected melanoma brain metastases, clinical outcomes, and responses to immunotherapy. JAMA Netw Open 6(8):e2329186

    Article  PubMed  PubMed Central  Google Scholar 

  36. Váraljai R, Horn S, Sucker A et al (2021) Integrative genomic analyses of patient-matched intracranial and extracranial metastases reveal a novel brain-specific landscape of genetic variants in driver genes of malignant melanoma. Cancers. https://doiorg.publicaciones.saludcastillayleon.es/10.3390/cancers13040731

    Article  PubMed  PubMed Central  Google Scholar 

  37. Hamilton R, Krauze M, Romkes M et al (2013) Pathologic and gene expression features of metastatic melanomas to the brain. Cancer 119(15):2737–2746

    Article  PubMed  CAS  Google Scholar 

  38. Mitra A, Andrews MC, Roh W et al (2020) Spatially resolved analyses link genomic and immune diversity and reveal unfavorable neutrophil activation in melanoma. Nat Commun 11(1):1839

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  39. Sperduto PW, Jiang W, Brown PD et al (2017) Estimating survival in melanoma patients with brain metastases: an update of the graded prognostic assessment for melanoma using molecular markers (melanoma-molGPA). Int J Radiat Oncol Biol Phys 99(4):812–816

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  40. Sperduto PW, Mesko S, Li J et al (2020) Survival in patients with brain metastases: summary report on the updated diagnosis-specific graded prognostic assessment and definition of the eligibility quotient. J Clin Oncol 38(32):3773–3784

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  41. Chen PL, Roh W, Reuben A et al (2016) Analysis of immune signatures in longitudinal tumor samples yields insight into biomarkers of response and mechanisms of resistance to immune checkpoint blockade. Cancer Discov 6(8):827–837

    Article  PubMed  PubMed Central  Google Scholar 

  42. Gopal YN, Rizos H, Chen G et al (2014) Inhibition of mTORC1/2 overcomes resistance to MAPK pathway inhibitors mediated by PGC1alpha and oxidative phosphorylation in melanoma. Cancer Res 74(23):7037–7047

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  43. Vashisht Gopal YN, Gammon S, Prasad R et al (2019) A novel mitochondrial inhibitor blocks MAPK pathway and overcomes MAPK inhibitor resistance in melanoma. Clin Cancer Res 25(21):6429–6442

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  44. Chen D, Barsoumian HB, Fischer G et al (2020) Combination treatment with radiotherapy and a novel oxidative phosphorylation inhibitor overcomes PD-1 resistance and enhances antitumor immunity. J Immunother Cancer. https://doiorg.publicaciones.saludcastillayleon.es/10.1136/jitc-2019-000289

    Article  PubMed  PubMed Central  Google Scholar 

  45. Najjar YG, Menk AV, Sander C et al (2019) Tumor cell oxidative metabolism as a barrier to PD-1 blockade immunotherapy in melanoma. JCI Insight. https://doiorg.publicaciones.saludcastillayleon.es/10.1172/jci.insight.124989

    Article  PubMed  PubMed Central  Google Scholar 

  46. Alvarez-Breckenridge C, Markson SC, Stocking JH et al (2022) Microenvironmental landscape of human melanoma brain metastases in response to immune checkpoint inhibition. Cancer Immunol Res 10(8):996–1012

    Article  PubMed  PubMed Central  CAS  Google Scholar 

Download references

Acknowledgements

We would like to sincere thank Dr. Jeffrey T. Chang from UTHealth McGovern Medical School at Houston for his mentoring, guidance, and critical input with the data analysis for SK. MAD is supported by the Dr. Miriam and Sheldon G. Adelson Medical Research Foundation, the AIM at Melanoma Foundation, the NIH/NCI P50CA221703, the American Cancer Society and the Melanoma Research Alliance, Cancer Fighters of Houston, the Anne and John Mendelsohn Chair for Cancer Research, and philanthropic contributions to the Melanoma Moon Shots Program of MD Anderson.

Funding

The work presented in this manuscript was fully supported by funds provided to MAD by the Dr. Miriam and Sheldon G. Adelson Medical Research Foundation, the AIM at Melanoma Foundation, the NIH/NCIP 50CA221703, the American Cancer Society and the Melanoma Research Alliance, Cancer Fighters of Houston, the Anne and John Mendelsohn Chair for Cancer Research, and philanthropic contributions to the Melanoma Moon Shots Program of MD Anderson.

Author information

Authors and Affiliations

  1. UT MD Anderson Cancer Center, Houston, TX, USA

    Swaminathan Kumar, Renato A. Guerrieri, Courtney W. Hudgens, Debora A. Ledesma, Fuchenchu Wang, Julie M. Simon, Kalman V. Katlowitz, Y. N. Vashisht Gopal, Jennifer L. McQuade, Lawrence N. Kwong, Jason T. Huse, Alexander J. Lazar, Jeffrey E. Gershenwald, Aron Y. Joon, Ken Chen, Ziyi Li, Prahlad T. Ram, Sherise D. Ferguson & Michael A. Davies

  2. Sarah Cannon Research Institute, Nashville, TN, USA

    Meredith S. Pelster

  3. The Ohio State University Comprehensive Cancer Center, Columbus, OH, USA

    Merve Hasanov

  4. Brigham and Women’s Hospital, Boston, MA, USA

    Grant M. Fischer

  5. Department of Quantitative Health Sciences, Division of Clinical Trials and Biostatistics, Mayo Clinic, Jacksonville, FL, USA

    Lauren E. Haydu

  6. University of California San Francisco, San Francisco, CA, USA

    Michael T. Tetzlaff

Authors
  1. Swaminathan Kumar
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Meredith S. Pelster
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Merve Hasanov
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Renato A. Guerrieri
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Courtney W. Hudgens
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Debora A. Ledesma
    View author publications

    You can also search for this author inPubMed Google Scholar

  7. Fuchenchu Wang
    View author publications

    You can also search for this author inPubMed Google Scholar

  8. Grant M. Fischer
    View author publications

    You can also search for this author inPubMed Google Scholar

  9. Julie M. Simon
    View author publications

    You can also search for this author inPubMed Google Scholar

  10. Lauren E. Haydu
    View author publications

    You can also search for this author inPubMed Google Scholar

  11. Kalman V. Katlowitz
    View author publications

    You can also search for this author inPubMed Google Scholar

  12. Y. N. Vashisht Gopal
    View author publications

    You can also search for this author inPubMed Google Scholar

  13. Jennifer L. McQuade
    View author publications

    You can also search for this author inPubMed Google Scholar

  14. Lawrence N. Kwong
    View author publications

    You can also search for this author inPubMed Google Scholar

  15. Jason T. Huse
    View author publications

    You can also search for this author inPubMed Google Scholar

  16. Alexander J. Lazar
    View author publications

    You can also search for this author inPubMed Google Scholar

  17. Michael T. Tetzlaff
    View author publications

    You can also search for this author inPubMed Google Scholar

  18. Jeffrey E. Gershenwald
    View author publications

    You can also search for this author inPubMed Google Scholar

  19. Aron Y. Joon
    View author publications

    You can also search for this author inPubMed Google Scholar

  20. Ken Chen
    View author publications

    You can also search for this author inPubMed Google Scholar

  21. Ziyi Li
    View author publications

    You can also search for this author inPubMed Google Scholar

  22. Prahlad T. Ram
    View author publications

    You can also search for this author inPubMed Google Scholar

  23. Sherise D. Ferguson
    View author publications

    You can also search for this author inPubMed Google Scholar

  24. Michael A. Davies
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

Each author listed as a co-author in this study have made substantial contributions in the following areas. SK, MSP, MH, RAG, CH, DL, FW, GF, JS, LEH, KK, VG, JM, LK, AYJ helped in the acquisition of data, helped with data analysis and interpretation of data. JH, AJL, MTT, JG, KC, ZL, PR provided input for data analysis and helped in interpretation of data. SK, SDF, MAD were involved in the conception, design of the work, interpretation of data, and substantively revised it. SK lead the entire data analysis, interpreted the results, drafted the work presented in the paper. MAD oversaw the entire work and provided mentorship to SK. All authors have reviewed the manuscript, approved the results and have provided input and comments wherever necessary.

Corresponding authors

Correspondence to Swaminathan Kumar, Sherise D. Ferguson or Michael A. Davies.

Ethics declarations

Ethics approval and consent to participate

MBM samples collected from patients who underwent craniotomy from 1991 to 2015 at MD Anderson Cancer Center was obtained from Central Nervous System (CNS) tissue bank and Melanoma Informatics, Tissue Resource, and Procurement Core (MelCore) facility under a protocol (LAB10-0519, titled “The Molecular Biology of Melanoma Brain Metastases”) approved by the Institutional Review Board.

Consent for publication

Not applicable.

Competing interests

MAD has been a consultant to Replimmune, Roche/Genentech, Array, Pfizer, Novartis, BMS, GSK, Sanofi-Aventis, Vaccinex, Apexigen, Eisai, Iovance, Merck, and ABM Therapeutics, and he has been the PI of research grants to MD Anderson by Roche/Genentech, GSK, Sanofi-Aventis, Merck, Myriad, Oncothyreon, Pfizer, ABM Therapeutics, and LEAD Pharma. MSP receives research funding paid to her institution Sarah Cannon Research Institute, Nashville, TN, United States by Abbvie, Actuate Therapeutics, Affini-T Therapeutics, Agenus, Arcus Biosciences, Astellas, BeiGene, BioNTech, Bristol-Myers Squibb, Codiak Biosciences, Compass Therapeutics, CytomX, Eisai, Elevation Oncology, Elicio, Exelixis, Fate Therapeutics, Fog Pharmaceuticals, Gilead, GlaxoSmithKline, HiberCell, Immune-Onc Therapeutics, Impact Therapeutics, Jazz Pharmaceuticals, Kura Oncology, Leap Therapeutics, Neogene, Novartis, OncXerna Therapeutics, Panbela Therapeutics, Revolution Medicines, Roche, SeaGen, SQZ Biotechnologies, Surface Oncology, Tachyon Therapeutics, Takeda, Translational Genomics, TransThera Sciences, ZielBio, 1200 Pharma. MSP also has a consulting and advisory role, where she receive payments paid to her institution, in Arcus Biosciences, AstraZeneca, CytomX, Daiichi Sankyo, Elevation Oncology, EMD Serono, Ipsen Biopharmaceuticals, Jazz Pharmaceuticals, Pfizer, SeaGen, Stemline Therapeutics, Takeda. Dr. Jason T Huse, an author in this manuscript, is also one of the Editors-in-chief in the Acta Neuropathological communications journal. No other authors disclosed any conflicts of interest to the study presented here.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

40478_2025_1978_MOESM1_ESM.pdf

Supplementary material 1: Figure 1. Immunohistochemistry analysis of CD3, CD8, PAX5 expressing cells in MBMs. Cell density of indicated cell populations in LTS (circles) and STS (triangles) MBMs. p < 0.05 was considered significant by Wilcoxon Rank Sum test. Figure 2. Kaplan–Meier analysis of post-operative OS of all patients included in this study. Median post-operative OS was 8.6 months for the 74 patients. For patients with RNA expression data from more than one surgically resected MBM, the earliest resection date was used for the analysis. Figure 3. Kaplan–Meier analysis of OS for MBM-only vs MBM-with_ECM patients. P-value was calculated using log rank sum test. p < 0.05 was considered significant.

40478_2025_1978_MOESM2_ESM.xlsx

Supplementary material 2: Table 1. LTS_vs_STS_upregulated_downregulated_gene_pathways_list (in excel sheet, Sheet 1) GENE_LIST. 164 differently regulated genes in MBMs from LTS compared to STS (8 downregulated genes and 156 upregulated genes in LTS MBMs). FDR adjusted P. value < 0.05 was considered significant. (in excel sheet, Sheet 2) GENE ONTOLOGY PATHWAYS. 1543 differently regulated gene ontology pathways in MBMs from LTS compared to STS (57 downregulated pathways and 1486 upregulated pathways in LTS). P.Up or P.Down < 0.05 was considered significant. (in excel sheet, Sheet 3) HALLMARK PATHWAYS. 12 differently regulated hallmark pathways in MBMs from LTS compared to STS (6 downregulated pathways and 6 upregulated pathways in LTS). FDR adjusted P. value < 0.05 was considered significant. (in excel sheet, Sheet 4) C2 CURATED GENE SETS. 574 differently regulated C2 curated gene sets in MBMs from LTS compared to STS (298 upregulated gene sets and 276 downregulated gene sets in LTS). FDR adjusted P. value < 0.05 was considered significant. (in excel sheet, Sheet 5) C7_IMMUNOLOGIC_SIGNATURES. 135 differently regulated C7 Immunologic gene signatures in MBMs from LTS compared to STS (98 upregulated gene signatures and 37 downregulated gene signatures in LTS). FDR adjusted P. value < 0.05 was considered significant.

40478_2025_1978_MOESM3_ESM.xlsx

Supplementary material 3: Table 2. BM_only_vs_BM_with_ECM_upregulated_downregulated_genes_pathways_list (in excel sheet, Sheet 1) GENE_LIST. 3 upregulated genes in MBM_site_only compared to MBM_with_ECM tumors. FDR adjusted P. value < 0.05 was considered significant. (in excel sheet, Sheet 2) GENE ONTOLOGY PATHWAYS. 67 upregulated gene ontology pathways in MBM_site_only compared to MBM_with_ECM. P.Up or P.Down < 0.05 was considered significant. (in excel sheet, Sheet 3) HALLMARK PATHWAYS. 9 differently regulated hallmark pathways in MBM_site_only compared to MBM_with_ECM (2 downregulated pathways and 7 upregulated pathways in MBM_site_only). FDR adjusted P. value < 0.05 was considered significant. (in excel sheet, Sheet 4) C2 CURATED GENE SETS. 287 differently regulated C2 curated gene sets in MBM_site_only compared to MBM_with_ECM (264 upregulated gene sets and 23 downregulated gene sets in MBM_site_only). FDR adjusted P. value < 0.05 was considered significant. (in excel sheet, Sheet 5) C7_IMMUNOLOGIC_SIGNATURES. 135 differently regulated C7 Immunologic gene signatures in MBM_site_only compared to MBM_with_ECM (98 upregulated gene signatures and 37 downregulated gene signatures in MBM_site_only). FDR adjusted P. value < 0.05 was considered significant.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Kumar, S., Pelster, M.S., Hasanov, M. et al. Integrated analysis of molecular and clinical features associated with overall survival in melanoma patients with brain metastasis. acta neuropathol commun 13, 75 (2025). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s40478-025-01978-1

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s40478-025-01978-1

Acta Neuropathologica Communications

ISSN: 2051-5960

Contact us