Brain metastases formation co-opts and reconfigures the brain microenvironment

The human brain was long considered an immune-privileged organ, and the role of the immune system in brain pathology was not given adequate consideration until evidence demonstrated the brain’s potential to elicit an immunogenic response1, which is suppressed by primary CNS tumors during progression2,3. Brain metastasis (BrM) frequently originates from disseminated melanomas, lung cancers, or breast tumors4. The tumor microenvironment (TME) at secondary sites largely regulates the growth of metastases from disseminated tumor cells. The significance of immune cells in the TME to therapy has been established by the success of immune checkpoint blockade5.

The brain TME is unique, containing cells not found outside the CNS. This includes but is not limited to astrocytes, cortical neurons, oligodendrocytes, and microglia that found in extracranial tissue at cell densities observed in the brain. This unique ecosystem exhibits limited inflammatory responses from infiltrating lymphocytes under normal physiological conditions and is remodeled by pro-tumorigenic macrophages in BrM6. The TME of BrM further differs from the ecosystem of cells in primary CNS tumors by including a more prominent population of molecularly diverse tumor-infiltrating leukocytes (TILs), less tumor-associated macrophages (TAMs), the presence of regulatory T-cells (Tregs), and the presence of plasma cells7,8. While the underlying mechanisms responsible for these differences are an active topic of investigation, the findings signify that BrM forms in part by modifying the brain’s immune milieu. This is supported by clinical evidence of increased survival among Glioma patients with increased monocytes and monocyte derived macrophages without a similar patient survival benefit from the presence of these two populations in BrM7. The implication of this observation is that these cells behave differently in BrM than primary CNS tumors. Likewise, transcriptomic analysis of flow-sorted CD45+ immune cells from tumors found a larger signature of differentially expressed genes in tumor-derived microglia from BrM than in tumor-derived microglia from primary CNS gliomas8. Further exploration in this study of the gene expression changes underlying these signatures found enrichment for type I IFN signaling in BrM-derived microglia but not in glioma-derived microglia. This is noteworthy given reports of a relationship between IFN signaling, immune suppression, and evasion of immune checkpoint blockade9. Conditioned media experiments demonstrating that secreted factors from resuspended BrM cells cause treated monocyte-derived macrophage cultures to upregulate genes in this signature8 further supports the notion of BrM reprogramming the immune components of the brain TME to permit the immune evasion neccesary for BrM growth.

Formation of BrM is a complex multistage process that co-opts diverse signaling pathways and is influenced by the intrinsic molecular subtypes of primary tumors, transcriptional and epigenetic changes accumulated by cancer cells, and the formation of a permissive microenvironment in the brain to seed metastases10. The primary tumor is understood to disseminate circulating tumor cells (CTCs) that traverse the blood-brain barrier (BBB), migrate by severing junctional adhesion molecule B (JAM-B) in response to Cathepsin S, and seed secondary tumors in the brain11. Once seeded, these cells are stimulated by signal transducer and activator of transcription 3 (STAT3), L1 cell adhesion molecule (L1CAM), and Cathepsin S signaling12,13. During this phase, the brain mounts an innate anti-tumor response by secreting plasmin from the brain stroma to transform the tumor necrosis factor (TNF) receptor family member Fas ligand (FasL), which triggers the caspase cascade and apoptosis14. CTCs seeding BrM overcome this by expressing serpins, which function as anti-plasminogen activators that inhibit plasmin and prevent FasL-induced apoptosis of the invasive cancer cells14,15. Reactive astrocytes equipped with STAT3 contribute to BrM formation by upregulating transforming growth factor α (TGF-α), epidermal growth factor (EGF), and macrophage migration inhibitor factor (MIF) to form astrospheres. Astrospheres are suppressive of CD8+ T cells in the brain, enabling immune evasion and growth of BrM12. The BrM niche is also primed for metastasis by molecular cues originating from extracellular vesicles (EVs) released from primary tumors and various tumor-associated stromal cells, including microglia, astrocytes, fibroblasts, and endothelial cells. These EVs play a significant role in facilitating the migration of cancer cells across the BBB and promoting brain tropism in addition to contributing to immune evasion16. Thus, understanding the key metastatic drivers and their crosstalk in the complex ecosystem of the brain TME is crucial for developing targeted immunotherapies and optimizing personalized medicine approaches to effectively and safely treat BrM (Fig. 1).

Fig. 1: The origin of brain metastases (BrMs) and their highly heterogeneous tumor microenvironment (TME), including the application of liquid biopsy as a diagnostic and prognostic tool in BrMs that allows better tumor management and more personalized therapeutic approaches.
figure 1

Circulating tumor cells (CTCs) are shed from various primary carcinomas or sarcomas into the bloodstream, after undergoing epithelial-mesenchymal transition (EMT). These tumor cells acquire the ability to cross the blood-brain barrier (BBB), responding to biochemical and mechanical cues that facilitate their invasion of brain tissue. Once in the brain, they interact with stromal cells, evading immune detection and promoting tumor progression through the expression of molecules that enhance metastatic behavior. Liquid biopsy technique, which isolates CTCs, circulating tumor DNA (ctDNA), extracellular vesicles (EVs), and non-coding RNA species from peripheral blood or cerebrospinal fluid (CSF), can improve the accuracy of BrM diagnosis and guide therapeutic strategies. These strategies may include antibody-based or cell-based immunotherapy (CAR T cells), chemotherapy, surgery, radiotherapy, targeted therapy, or combinations of these approaches.

Full size image

Adaptation to the Brain Confers a CNS-Specific Phenotype to Brain Metastases

Secondary site-specific colonization and immune evasion lead to systemic and CNS metastases possessing different immune profiles and genetic signatures when compared to the primary tumor, including differences in mutations and heritable epigenetic modifications16. To communicate within the TME and cross the BBB, metastatic cancer cells express a set of genes that facilitate BrM formation under hypoxic and hypoglycemic conditions17. Transcriptome profiling in situ using an animal model of early metastatic colonization by breast cancer cells reveals that these expression programs are secondary site-specific, with cells seeding BrM demonstrating lower levels of oxidative stress and antioxidative response relative to cells forming lung metastases and mammary tumors18. These results suggest adaptation and/or selection of BrM cells for fitness under the unique metabolic conditions present in the brain19, a prospect supported by profiling of patient specimens demonstrating that metastatic brain lesions harbor mutations distinct from those present in the same patient’s primary tumor, regional lymph nodes, bloodstream, and extracranial metastases20.

In addition to metabolic adaptation, cancer cells seeding BrM depend heavily on interactions with endothelial cells to metastatically colonize the brain14. An elegant study by Karimi et al.21 highlights the role of differential gene expression in microvascular proliferation and the expression of tight junction protein claudin 5, which increases vascular permeability in BrM. Their findings reveal that cancer cell-associated endothelial cells show increased expression of the proliferative biomarker Ki67 in both the core and margins of BrM, indicating multiple foci of aggressive pro-angiogenic tumor cells. Moreover, the authors demonstrate the suppression of endothelial cell proliferation in BrM occurs via CD8+ T-cells interactions, suggesting that the immune system plays a role in limiting BrM progression by constraining tumor angiogenesis. This is particularly noteworthy given the established role of the vascular basement membrane (VBM) in BrM as a pro-growth substrate or ‘niche’ for disseminated tumor cells that supports their adhesion and provides proliferative signaling22.

Tumorigenesis under these conditions produces BrM that are biologically distinct from both the primary tumor that seeds BrM and their sibling metastases in extracranial organs. These differences are partly attributable to intrinsic molecular features enriched among CTCs that survive and successfully colonize the brain. However, the extrinsic influence of the neurovascular, astrocytic, and immune components of the brain TME is considerable.

Diverse Immunological Signatures of BrMs Based on the Primary Tumor Type

The degree of BrM-associated inflammation varies by primary tumor, with the most prominent inflammation observed in BrMs originating from melanoma, and less severe inflammatory responses in BrMs originated from breast carcinomas21. Melanoma BrMs are particularly rich in monocytes, microglia, and accumulated CD8+ T-cells compared to lung and breast BrM. Tumor-infiltrating macrophages (TIMs) constitute the main component of the BrM immune signature, regardless of the primary tumor of origin, followed by infiltrating T-cells. Resident myeloid cells and microglia are also generally present in low numbers in lung and skin BrMs, but microglial infiltration is almost absent in breast BrM23. Furthermore, the TME of BrM has significantly higher levels of TILs compared to primary glioblastoma (GBM), which are considered immunologically ‘cold’. This suggest a higher priming rate for anti-tumor T-cell responses in tumors of extracranial origin, likely due to the higher mutational burden found in melanoma and lung cancer. BrMs may also attract more T cells due to the release of immunomodulatory chemokines and activation of pathways that promote T-cell trafficking and extravasation into these secondary tumors24. Despite the increased T-cell numbers in BrM, there is limited T-cell-mediated anti-tumor response due to the upregulation of immune checkpoint molecules such as PD-1, LAG-3, and TIGIT. Therefore, analyzing the presence of inhibitory receptors in BrM in the context of the primary tumor type is essential to optimizing the application of immune checkpoint inhibitors (ICI), especially for melanoma25.

Spatial analysis demonstrates an association between the distribution of extracellular matrix (ECM) components and the preferential localization of different immune cells to ECM-rich regions26. TIMs serve as the major immune cell type influencing brain TME remodeling across both stromal and neoplastic cell (NC)-rich regions27. analyzed the leukocyte landscape of brain tumors using high-dimensional single-cell profiling (CyTOF) and revealed differences between tissue-resident and invading immune cells within the TME of gliomas and BrMs. The glioma TME predominantly featured tissue-resident reactive microglia, whereas BrMs were enriched with accumulated TILs.

Recently, Wischnewski et al.28 used transcriptomics to elucidate the composition and role of T-cells in both primary and BrM tumors. Their analysis of circulating and tumor-infiltrating T-cells from 84 individuals with brain cancer (36 with GBM and 48 with BrM) along with 44 individuals with extracranial tumors originating from lung and breast cancer, identified a subgroup of BrM patients with clonally expanded, potentially tumor-reactive T cells (pTRT)28. The study also found that CXCL13+CD39+CD8+ TILs were enriched in pTRT cell-high BrM, similar to primary non-small cell lung cancer NSCLC29. The authors propose that individuals with pTRT cell-high BrM may benefit from immune checkpoint blockade, paving the way for individualized ICI treatments based on immunogenic expression profiling29.

Other factors that may influence the choice of ICI include the abundance of CXCL13-expressing (CD39+) CD8+ T cells and the expression of CXCL9 in the TME, both of which are predictors of immunotherapy response in several primary malignancies, including lung and triple-negative breast cancer30. Myeloid cells in the TME may also impact ICI susceptibility. In a study of primary human gliomas, bone marrow-derived myeloid cells (BMDMs) were distinguished from brain-resident microglia using CD49D positivity, showing an association between BMDM infiltration and increased immunosuppression in analyzed tumors31. A separate study by Guldner et al.32 found that human lung, breast, and skin BrMs possess myeloid cells recruited both locally from the brain parenchyma and from BMDMs. Their experiments using a murine BrM model further demonstrated a role for myeloid cells of both origins in promoting BrM through immunosuppression.

Together, these findings suggest that tumor resident T-cells in BrM are primed for an anti-tumor response but are rendered inactive due to the effects of other TILs within the TME. Nevertheless, BrMs, particularly those originating from melanoma, have shown vulnerability to ICI,. An open-label, multicenter, phase 2 study with Nivolumab plus Ipilumab demonstrated clinical benefits, with stable or reduced intracranial disease in 57% of BrM patients33. More recently combined ICI targeting both LAG-3 and PD-1 using Relatlimab in combination with Nivolimumab increased progression-free survival to nearly 48%, compared t 36% with Nivolimumab monotherapy, in a Phase II/III study of melanoma BrM pateints34.

Breast cancer BrM

Human epidermal growth factor receptor 2 positive (HER2-positive) and triple-negative (estrogen receptor/progesterone receptor/HER2 negative, TNBC) breast cancers are the most aggressive molecular subtypes of this disease, with TNBC having the lowest overall survival rates35. TNBC exhibits high rates of BrM, with estimates suggesting that up to one-third of women third of women diagnosed with TNBC will develop BrM36. Pathogenic BRCA1 mutations in TNBC further predispose its progression to BrM37. A comparison of gene expression profiles of primary breast cancer metastasizing to the brain and those metastasizing to extracranial organs reveals that T cell activation is more prominent in BrM-forming primary tumors. T lymphocytes promote breast cancer cell invasion and metastasis to the brain by increasing vascular permeability, thus facilitating passage across the BBB. These effects have been attributed to the increased transcription and translation of guanylate-binding protein 1 (GBP1)38. Additionally, significant downregulation of MHC class I antigen presentation proteins has been observed on metastatic TNBC cells, suggesting that their metastatic potential is linked to immune evasion39.

Giannoudis et al.40 characterized the immune microenvironment of primary breast cancers and compared it to that of breast cancer BrM. Their study reveals a significant reduction of TILs in BrM compared to primary breast cancers. The authors also identified changes in 112 genes, including those involved in ECM remodeling, cytokine and chemokine signaling, lymphocyte regulation, antigen presentation, and immune cell adhesion and migration. Notably, PD-L1 (CD274), CTLA4, TIGIT and CD276 (B7H3) were downregulated in most BrMs, except in those originating from TNBC, where higher PD-L1 and CTLA4 expression were reported. CTLA4 expression was also reported to be higher in HER 2-positive versus estrogen receptor-positive breast cancer BrM. Increased ARG2 expression was also observed in breast cancer BrM.

These findings demonstrate that metastatic breast cancer exhibits an immune TME distinct from that primary breast cancer. BrM from breast cancer also display heterogeneity in the expression of immune modulators and conventional immune checkpoint targets, making it difficult to pinpoint a singular approach for immunotherapeutic targeting of these metastases. A personalized approach using ICI in combination with other therapeutic modalities may offer the most promising strategy for targeting breast cancer BrM.

Lung cancer BrM

Approximately 16–60% of NSCLC patients develop BrM41. Recent single-cell transcriptomic analysis by Wang et al.42 characterized the TME of NSCLC BrM as a remodeled, immunosuppressive, and fibrogenic niche that supports metastatic growth. This is accomplished by a combination of decreased antigen presentation by tumor cells and TAMS, diminished B and T cell activation, alongside an increase in neutrophils, M2 macrophages, and reactive astrocytes. The authors also identified disruptions in fibrosis and immune regulation as common features between the primary lung tumor and BrM TMEs formed by NSCLC.

Using bulk tumor DNA profiling and immunohistochemical (IHC) methods, Song et al.43 also analyzed primary and BrM NSCLC. Their study found a higher tumor mutational burden (TMB) in primary NSCLC compared to BrM, which may reflect the early dissemination of metastatic clones, with the primary tumor continuing to accumulate mutations at higher rates. The authors also detected increased oligoclonal T-cell expansion by T-cell receptor repertoire (TCR) sequencing and equivalent PDL1 expression by IHC in BrM compared to primary NSCLC. Notably, differences in co-mutations were identified between EGFR-mutant BrM and EGFR-WT BrMs, with immune-related pathways enriched in EGFR-mutant BrMs. The study further demonstrated that increased CD8+ TILs in NSCLC BrM predicted better overall survival following brain surgery. This association, along with Wang et al.42 findings on immune dysregulation in the BrM TME, provides hope that NSCLC BrMs may respond to ICI. However, the differences observed in BrM arising from distinct NSCLC mutational subtypes support the need for personalized ICI approaches within the lung BrM patient cohort.

Renal cell carcinoma BrM

Renal cell carcinoma (RCC) is an immunogenic tumor that accumulates dysfunctional immune cell infiltrates, which fail to control tumor growth effectively44. Between 25–30% of patients progress to develop metastatic RCC after treatment of their primary tumor45. The median overall survival (OS) for untreated patients with RCC BrM is poor, averaging between 3-4 months46.

As demonstrated by Braun et al.44, RCC BrM has significant infiltration of TILs, consistent with the tumor’s immunogenic nature. However, the immunosuppressive characteristics of metastatic RCC cells attenuate the function of effector T-cells residing in the BrM TME38. Using single-cell RNA and T-cell receptor sequencing, the authors identified an enrichment of terminally exhausted CD8+ T cells, M2-like immunosuppressive macrophages, and the expression of several ligand-receptor pairs by the infiltrating leukocytes that support T-cell suppression. In primary RCC, co-expression of PD-1 and Tim-3 on TILs has been associated with advanced disease stages and poor clinical outcomes47. Additionally, immunosuppressive cytokines and stromal cell secretions have been reported to promote T cells exhaustion and impair cytotoxic function against cancer cells48. These findings provide a rationale for combination approaches to ICI therapy in RCC BrM, with a focus on addressing changes in the TME that greatly contribute to immunotherapy resistance.

Melanoma BrM

Metastatic melanoma evades detection and elimination by the immune system by manipulating TILs. Tumor cell secretion of TGFβ activates immunosuppressive Tregs and the enrichment of activated CD8+ T-cells with upregulated PD-L1 expression in the TME suppresses the anti-tumor immune response49. While soluble PD-L1 has been identified as a biomarker for melanoma, its expression being limited to certain melanoma subtypes restricts its broad application across all melanomas50. Emerging biomarkers of anti-tumor immunity in melanoma metastases include measures such as cytotoxic T-cell turnover and tumor infiltration by CD8+ memory effector T-cells51. Additionally, novel molecular screening tools are distinguishing between naïve, early, and late dysfunctional states of intratumoral CD8+ T-cells, which may prove valuable in guiding decisions regarding immunotherapies for patients with metastatic melanoma52.

A recent study expanded these insights by deploying single-cell analysis to characterize the immune microenvironment of human melanoma BrM in response to immune checkpoint inhibition53. The authors observed heterogeneity in the myeloid compartment of the melanoma BrM TME, including distinct neutrophil states, such as IL8-expressing population that correlated with tumor cell epithelial-to-mesenchymal transition (EMT). Additionally, the phenotype, clonotype, and overall quantity of infiltrating T cells in melanoma BrM were associated with response to ICI. The study also reported an expansion of exhausted and cycling T-cell clones, alongside non-clonally expanded CD4 + FOXP3+ T cells, memory CD8+ T cells, and IFN-responsive CD8+ T cells. Phenotypic differences between blood-overlapping and tumor-exclusive T-cell clones implied crosstalk between peripheral and intratumoral immune compartments in melanoma BrM.

It is generally agreed that an effective strategy for targeting melanoma BrM involves converting the brain’s immunologically ‘cold’ environment into an immunoreactive TME54,55. A retrospective analysis of more than 2700 patients with stage 4 melanoma enrolled in the US National Cancer Database between 2010 and 2015 who presented with BrM supports this approach, showing that patients receiving ICI had a of median survival of 12.4 months, compared to to 5.2 months for those receiving other treatments56. Additional avenues for transforming the melanoma BrM TME from cold to hot include the development of antibodies targeting immunosuppressive cytokines like TGF-β, or IL-1055. This could be particularly relevant in melanoma BrM, given the known role of elevated TGF-β2 in the expansion of immunosuppressive Tregs57 and preclinical evidence supporting this approach in murine models of melanoma BrM58.

Colorectal cancer BrM

BrM in patients with colorectal cancer (CRC) are infrequent and portend a poor prognosis when they occur59,60. Studies with small sample sizes have identified various mutations in PIK3CA and BRAF61, overexpression of EGFR62 and CXCR463, methylation of MGMT64, and elevated CA19-965 and carcinoembryonic antigen (CEA) levels as risk factors associated with CRC BrM66,67. Furthermore, correlations between KRAS mutations and BRAF mutations, as well as the role of chemokine receptor type 4 (CXCR4) have been identified in the context of BrM63. These findings suggest that the accumulation of heterogenous tumor cell-intrinsic events in primary CRC may play role in determining the rare seeding of BrM, a notion supported by a recent study of 17 metastatic CRC (mCRC) patients, which found that most metastases arose from independent subclones within the primary tumor68.

In another study, Sun et al.69 have also demonstrated mutations in CRC BrM that are distinct from those present in primary CRC in a cohort of 19 mCRC patients, with homologous recombination deficiency (HRD) and mismatch repair deficiency (MMRd) mutations elevated in BrM but not in respective primary tumors. Similarly, Roussille et al.70 evaluated CRC BrM in comparison to primary CRC tumors and found lower levels of immune infiltrates in BrM. This aligns with findings from Harter et al.71, who reported low tumor expression of PD-L1+ and PD-1+ tumors alongside low rates of CD3+ and CD8 + T-cells in CRC BrM samples. These results highlight a consistent pattern observed across BrM from different tumor types. BrM are more immunologically cold than their respective primary tumors. As such, methods aimed at increasing immune reactivity in the TME of CRC BrM may confer clinical benefit to patients. However, limited data exist on the use of ICIs in CRC BrM due to the lack of adequately powered clinical studies to assess their efficacy in this setting.

Sarcoma BrM

Sarcomas are a molecularly and phenotypically diverse group of malignancies that are rare, can develop in any organ, and arise from the neoplastic progression of mesenchymal cells72. While outcomes are generally favorable for patients with resectable organ-confined disease, those presenting with metastases have a median overall survival of only 18 months73. Combination chemotherapy and signal transduction inhibitors offer initial clinical benefits in patients with metastatic sarcoma who have the appropriate performance status and relevant tumor mutations, but acquired resistance remains a persistent treatment challenge74. Sarcoma BrM are rare buth highly lethal, occurring in 1-8% of sarcoma patients, with mortality typically within 1 year of diagnosis73. Given that sarcoma BrM patients are usually oligometastatic, systemic therapy remains a mainstay75 despite the poor CNS penetration of many agents, which is considered to contribute to the increased incidence of BrM in heavily treated metastatic sarcoma patients76.

The potential for immunotherapy to provide a therapeutic breakthrough for patients with sarcoma BrM is still uncertain. Both tumor cells and tumor-infiltrating white blood cells in primary sarcomas express immune checkpoint molecules, particularly PD1/PD-L177. The observation has led to a series of clinical trials in both primary and metastatic sarcoma patients, including trials combining VEGF inhibitors with ICIs, given the known role of VEGF in suppressing immune response in sarcomas78. Although deep single-cell profiling of immune infiltrates in sarcoma BrM, similar to studies in NSCLC and melanoma BrM are not currently available, correlative studies from clinical trials suggest a role for PD-L1-expressing macrophages, tumor resident Tregs, and tumor resident memory effector T cells in predicting immunotherapy response in metastatic or unresectable sarcomas77. Given the rarity of sarcoma BrM within an already uncommon tumor type, powering deeper studies to identify biomarkers for personalized immunotherapy will likely require a consortium-scale effort.

Liquid biopsy as a powerful tool for BrM prognosis and treatment management

While not the core focus of this review, it is worth mentioning that due to the high heterogeneity of BrMs, the application of sophisticated cancer care tools such as liquid biopsy provides a more accurate assessment of cancer progression and patient outcomes78. Several recent studies in solid tumor management imply that this technique substantially improves early metastasis diagnosis and supports precision and personalized medicine approaches, especially in tailoring immunotherapeutic strategies79,80,81. As a diagnostic and prognostic tool in oncology, liquid biopsy allows the evaluation of different body fluids at cellular and molecular levels to differentiate and stratify biomarker candidates associated with metastasis. In the context of BrM, peripheral whole blood and cerebrospinal fluid (CSF) are the two most valuable fluids for analysis79,82.

Liquid biopsy typically analyzes CTCs, circulating tumor DNA (ctDNA), microRNAs (miRNAs), long non-coding, RNAs (lncRNA), circular RNAs (cirRNA), and tumor-associated EVs81. Blood-derived CTCs provide a valuable source for identifying BrM-specific biomarkers. Nevertheless, as these cells disseminate from primary tumor sites, they often acquire novel mutations and distinct epigenetic signatures that in turn determine their metastatic organotropism79. Consequently, these alterations may lead to either dormancy or the development of specific immune evasion mechanisms at new metastatic sites in the brain, including upregulation of CXCL8, CXCR4, and CD86, as demonstrated by Herbst et al.83. On the other hand, CTCs isolated from CSF seem to have a unique molecular profile compared to CTCs from peripheral blood or extracranial CTCs84, making them more informative for therapeutic management and patient stratification85.

Additionally, ctDNA isolated from patients with BrM possesses specific methylation patterns not identified in other primary or secondary tumor sites, and these patterns such as EGFR variants could serve as primary prognostic markers86. An illustrative study by Zuccato et al.87 working with lung adenocarcinoma patients developed a glmnet classifier informed by methylation signatures from cell-free DNA in the plasma of patients with intracranial masses that span the entities commonly considered in a differential diagnosis: glioma, CNS lymphoma, and brain metastasis. Their classifier accurately discerned with an AUC of 0.8 when plasma samples were from BrM patients in a validation set that included glioma, CNS lymphoma, and non-CNS control cases. The group achieved median accuracy of 74.2% when the classifier was tested in a second cohort. It is noteworthy that this performance was achieved using solely methylome analysis of liquid biopsies, highlighting the potential for this method as non-invasive diagnostic tool. In addition to DNA, differential expression of non-coding RNA species, particularly miRNAs (e.g., those targeting PTEN) and lncRNAs (e.g., XIST downregulation), shows significant potential as a diagnostic aid in BrM88,89.

Lastly, EVs are potent modulators of the BrM TME. Thus, their isolation and high-resolution molecular characterization hold great promise for managing secondary cancers in the brain. For instance, Serrati et al.90 recently showed that PD-L1-positive exosomes derived from metastatic melanoma cells induced resistance to anti-PD-1 immunotherapy. Although challenges remain in the exploitation of liquid biopsy for BrM management, this method, together with advanced molecular analyses such as genomic sequencing and epigenetic profiling, offers an invaluable and complementary tool for understanding the complex molecular landscape of BrM. A deeper understanding of these mechanisms could enable the development of personalized therapeutic strategies, ultimately leading to improved patient outcomes.

Immune checkpoints and ICIs

As reviewed above, co-opting immunosuppressive mechanisms to establish and maintain tumors is a common theme across BrMs originating from various primary malignancies. This is frequently achieved by exploiting immune checkpoint biology. In some cases, metastatic tumor cells express the necessary ligands, while in others, infiltrating cells within the BrM TME perform this function. Regardless, ICIs disrupt interactions between checkpoint ligands and their receptors on effector T cells, restoring CD8+ T-cell-mediated clearance of tumor cells in BrM. The sections below describe current ICI by the specific checkpoint targeted. This list will likely grow as additional checkpoints and strategies for targeting them are developed (Fig. 2).

Fig. 2: Historical timeline of FDA-approved immunotherapies used in the treatment of brain metastases (BrMs) over the last 13 years.
figure 2

The monoclonal antibodies (mAbs) are color-coded based on their targets: PD-1 mAbs are marked in yellow, PD-L1 in green, LAG-3 in red, and CTLA-4 inhibitors in blue.

Full size image

Anti-PD-1 Immunotherapies

The programmed cell death protein-1 (PD-1)/programmed death-ligand 1 (PD-L1) checkpoint regulates the induction and maintenance of immune tolerance within the TME. Activation of PD-1 on T cells by its ligands, PD-L1 or PD-L2, impairs activation, proliferation, and cytotoxic activity necessary for an effective anti-tumor response. Nivolumab, a fully human monoclonal antibody (mAb) against the PD-1 cell surface receptor, inhibits its interaction with PD-L1. It has received FDA approval for several indications including metastatic melanoma, disease progression following Ipilimumab therapy, and metastatic squamous NSCLC with progression on or after platinum-based chemotherapy79. A key advantage of Nivolumab is its durable response, observed in as many as 68% of patients who meet specific predictive serum biomarker criteria80. Tables 1–3 list the clinical trials that have investigated the use of Nivolumab in BrMs.

Table 1 Outcomes of immune checkpoint inhibitors in the treatment of NSCLC with BrM
Full size table
Table 2 Clinical trials with renal cell carcinoma and brain metastasis
Full size table
Table 3 Clinical trials with immunotherapy in patients with melanoma and brain metastasis
Full size table

Pembrolizumab, another humanized IgG4 mAb targeting PD-1, recently received its 40th FDA approval for advanced endometrial cancer patients based on clinical trials supporting its safety and efficacy81. This mAb is also used in solid tumors, including BrM and primary CNS malignancies, that have high microsatellite instability (MSI-H), DNA mismatch repair deficiency (dMMR), or high tumor mutational burden (TMB-H) based on data across 30 cancer types from the KEYNOTE trials (KEYNOTE-051, -158, and -164) [NCT02332668, NCT02628067, NCT02460198]. Another PD-1 inhibitor, Cemiplimab, is currently under limited investigation in patients with BrM.

In an open label phase 2 study by Goldberg et al.91 on NSCLC and melanoma patients with BrM, 29.7% of patients with PD-L1 expression of atleast 1% demonstrated BrM response to Pembrolizumab. It is noteworthy that PD-L1 expression was the only stratifying criteria in this study, suggesting a role for MSI, dMMR, and TMB in improving response rates. An independent study also investigated the value of optimizing assessments for the proportion of the tumor expressing PD-L1 (TPS score), finding a 54.7% response rate to ICI regimens using pembrolizumab among NSCLC BrM whose tumors had a median TPS of 7.5%92. The study demonstrated that patients with higher PD-L1 expression (TPS > 40%) had significantly longer intracranial progression-free surivival, suggesting a role for checkpoint ligand expression as a predictor of both therapeutic response and response duration for patients receiving PD1 targetted ICI. A challenge when implementing PD-L1 expression as a predictive biomarker is the need to assess expression in BrM at the start of therapy. Analysis of 73 primary-BrM pairs of lung cancer found discordance in tumor cell expression between primary and BrM specimens, with most discordant pairs having biopsies that were obtained more than 6-month apart93. The observation both discourages the use of the primary tumor for such assessments and highlights the need to account for temporal heterogeneity. Spatial heterogeneity presents a similar challenge, as tumor cells expressing PD-L1 with no nearby CD8+ T-cells will be unemcumbered by PD1 targetted ICI. Immunoscore-IC is an in vitro diagnostics assay that addresses this by computing a score that accounts for both the densities of and distance between CD8+ T-cells and PD-L1+ tumor cells and performs well as a predictor of progression-free survival in metastatic NSCLCs patients treated with anti-PD-1 or anti-PD-L1 therapy94. The study was not limited to BrM patients, but nevertheless illustrates the need to account for complexity in the BrM TME when developing response biomarkers for ICI.

PD-L1 inhibitors

PD-1 has two ligands, PD-1 ligand 1 (PD-L1) and PD-1 ligand 2 (PD-L2). The expression of PD-L1, whether by immune cells in the TME or tumor cells, serves as a predictive biomarker for response to therapies targeting the PD-1/PD-L1 checkpoint82. Beyond PD-1, PD-L1 can also inhibit T-cell activation by binding to the negative regulator CD80 (B7.1)83. Therefore, PD-L1’s effect on T cells includes molecular processes beyond what PD-1 inhibition alone addresses. Antibodies targeting the PD-L1, such as Adrebrelimab, Atezolizumab, and Durvalumab, may address this concern. A recent single arm phase 2 study of Atezolizumab in combination with chemotherapy established median overall survivial of 11.8 months and 2-year overall survival rate of 27.5% in NSCLC patients with previously untreated BrM on this regimen95. While the single-arm design of this study limits assessment of Atezolizumab’s contribution to survival in this cohort of combined chemotherapy-treated patients, the regimen was well tolerated with an overall response rate of 45% and median progression-free survival of 8.9 months.

CTLA-4 inhibitors

Cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) is expressed by T cells, and constitutively present on the surface of Treg cells. Upon activation by TCR or CD28 signaling, it is induced to localize on naïve T cells. CTLA-4 binds B7-1 and B7-2 on antigen-presenting cells (APCs), leading to T cell anergy. Tumors expressing B7-1 and B7-2 exploit this mechanism to suppress T-cell84. mAbs targeting CTLA-4, such as Ipilimumab, interfere with this interaction and have shown efficacy in clinical trials of melanoma85. Ipilimumab, an IgG1κ anti-CTLA4 mAb, gained FDA approval in 2011 for non-resectable stage III/IV melanoma. It has been extensively tested in patients with melanoma BrM and NSCLC BrM, with success observed in combination with Nivolumab86.

Anti-lymphocyte activation gene-3 antibody

Lymphocyte Activation Gene-3 or LAG-3 (CD223) is an inhibitory receptor that upregulates during physiologic immune responses to infections, serving as a checkpoint to prevent tissue damage and auto-immunity. LAG-3 binds to MHC class II molecules with greater affinity than CD4, attenuating cytotoxic T-cell responses. Tumor cells and the TME in BrM often co-opt this mechanism by overexpressing LAG-388. Anti-LAG-3 therapies include LAG-3-targeting mAbs, bispecific LAG-3 antibodies, and fusion proteins. These immunotherapies suppress anti-inflammatory cytokines such as IL-12 and IFN-γ89. Clinically, LAG-3 antibodies are used in combination with PD-1 inhibitors, as LAG-3 and PD-1 are co-expressed on tumor-infiltrating CD8 + T-cells. Relatlimab is the first LAG-3 inhibitor approved by the FDA for use in combination with Nivolumab, for treating unresectable or metastatic melanoma based on the RELATIVITY-047 trial34. Another LAG-3 antibody, Fianlimab, is currently being tested in combination with Cemiplimab in a phase III trial (NCT05352672) for patients with previously untreated locally advanced or metastatic melanoma.

Combinatorial approaches to immune checkpoint inhibition

Combinatorial approaches involving LAG-3 inhibitors, such as Relatlimab, in combination with PD-1 inhibitors like Nivolumab, have shown promising outcomes in advanced melanoma, as demonstrated by improved progression-free survival in the RELATIVITY-047 trial34. While this trial did not focus specifically on brain metastases, its results highlight the potential efficacy of dual checkpoint blockade in immunologically suppressed tumor microenvironments, such as those found in brain metastases. Ongoing clinical trials, such as NCT05352672 evaluating Fianlimab with Cemiplimab in advanced melanoma, will further inform on the potential of dual checkpoint inhibition strategies to optimize immunotherapy responses. The rationale for combining LAG-3 and PD-1/PD-L1 blockade lies in the co-expression of these pathways on exhausted CD8+ T cells, where dual checkpoint inhibition can effectively reinvigorate T-cell function96.

While these registrational phase 3 trials are not specific to brain metastases, a multi-center open label phase 2 study by Long et al.97 focused-on melanoma brain metastasis evaluated therapeutic responses to the combination of nivolumab and ipilimumab and to monotherapy with nivolumab alone. The study found intracranial responses in 46% of patients in the combination therapy arm, nearly twice the 20% response rate to nivolumab monotherapy. Although this study has considerably smaller enrollment than the phase 3 studies mentioned, it highlights the extent to which therapeutic benefits of combination ICI reported by registrational studies that are not enriched for BrM patients may nevertheless carry-over when tested in this setting.

Immune checkpoint inhibition across tumors of differing primary origin

More granular insights into tumor microenvironment (TME) adaptations across different primary tumor origins, such as breast cancer and renal cell carcinoma, reveal significant heterogeneity in immune composition and suppressive mechanisms98. In breast cancer BrM, particularly from triple-negative and HER2-positive subtypes, reduced T-cell infiltration and upregulation of immune checkpoint molecules such as PD-L1 and CTLA-4 are observed, contributing to immune evasion99. In contrast, renal cell carcinoma BrM are characterized by infiltrating exhausted CD8+ T cells and M2-like macrophages, creating a profoundly immunosuppressive environment100. Aligning therapeutic strategies, such as immune checkpoint inhibitors or combination therapies, with these tumor-origin-specific immune adaptations could further optimize immunotherapy responses and improve clinical outcomes.

Targeting BrMs with chimeric antigen receptor T-cell therapy

Over the past decade, researchers have developed chimeric antigen receptor (CAR) T-cell therapy through various innovative methods and techniques, including leukapheresis, immunomagnetic enrichment, genetic engineering, and ex vivo T-cell expansion90. This immunotherapy involves modifying patients’ T effector cells to target malignant cells via their specific epitopes, such as CD19 and BCMA, primarily in hematological cancers, thus significantly suppressing tumor growth101.

Since CAR T-cell therapy is a cell-based approach, its native form is autologous, using the patient’s own T cells to minimize the risk of immune rejection. However, cutting-edge technologies like induced pluripotent stem cells (iPSCs) are now opening new avenues towards scalable, allogeneic ‘off-the-shelf’ immune cell therapeutics. These cells are engineered to be non-immunogenic through specific genetic modifications during their manufacturing process102,103. Furthermore, these artificial iPSC-derived cells could be genetically engineered to overcome tumor resistance, enhance on-target efficacy, and reduce off-target cytotoxicity in situ104. Despite its success in hematologic malignancies, CAR T-cell therapy for solid tumors, particularly complex malignancies such as BrMs, is currently restricted in scope due to several key obstacles105. These include antigen heterogeneity and epitope downmodulation, on-target off-tumor cytotoxicity and related safety concerns, poor trafficking across the BBB and reduced intratumor infiltration, T cell exhaustion, and immunosuppressive TME106. As discussed earlier, the TME of BrMs is highly heterogeneous, with extensive antigenic variation often depending on the primary tumor type and the mutation status of distinct CTCs. Therefore, selecting appropriate BrM-specific antigens with sufficient antigen density for CAR recognition is crucial for the effective eradication of secondary malignancies while preventing tumor-induced immunosuppression and minimizing cytotoxicity to local healthy tissue107,108. A recent preclinical study using a human xenograft model demonstrated the effective application of HER2-BBζ CAR T-cell immunotherapy against multifocal breast carcinoma metastasis to the brain, leading to significant tumor regression105. An important aspect of its efficacy was the route of administration. Regional intraventricular delivery significantly enhanced T-cell persistence and likely supported antigen-dependent trafficking, whereas systemic approaches showed a significant reduction in T-cell anti-tumor activity105,109.

In conclusion, cell-based immunotherapy, particularly allogeneic CAR T-cell therapy, holds great promise for the targeted and efficient reduction of solid tumors, including BrMs. Local injection of cell therapeutics into the brain appears to be crucial for overcoming the challenges posed by the BBB and/or poor cell trafficking within the hostile BrM TME. Complementarily, more efficient screening pipelines for antigen candidates and advanced engineering of CAR constructs could be key to mitigating on-target, off-tumor toxicity106.

Perspectives

TME analysis of different types of cancer and metastasis to the brain has revealed several promising biomarkers and therapeutic targets using state-of-the-art techniques such as single-cell sequencing technologies, multiplex imaging, spatial profiling, and bulk tumor profiling. In the case of BrM, a common theme across the spectrum of primary tumors of origin is a suppressed immune response in the BrM TME and a mutational profile unique to BrM when compared to the primary and extracranial “sibling” metastases. Analysis of patient cohorts with matched primary tumor samples also reveals differences in biomarker profiles for PD-1, PD-L1, and tumor-infiltrating CD8 + T cells.

Data on clinical response rates to ICI in patients with BrM are limited, as progressing to BrM is often an exclusion criterion for larger clinical trials evaluating the efficacy and safety of ICIs. However, smaller studies show efficacy in the form of intracranial tumor necrosis of BrM after only a single cycle of ICI treatment combined with chemotherapy in patients with NSCLC BrM110. Findings such as these call for study designs that include BrM patients in ICI efficacy trials and nest validation of predictive and response biomarkers. Molecular information such as tumor mutational burden, mismatch repair system status, and PD-L1 expression, in addition to the cellular density of tumor-infiltrating CD8 + T-cells, should also be collected in BrM patients to facilitate better prediction of response to ICIs by intracranial tumors. Additionally, understanding the impact of mutations in BrM that differ from those present in the primary tumor on BrM-specific tumor biology will prove essential to adequately caring for patients with this lethal diagnosis.

A recent meta-analysis of 56 studies with 2739 patients has shown a high level of disparity in PD-L1, PD-1, PD-L2, and TIL levels alongside a lesser likelihood of disparity in the tumor mutational burden and microsatellite instability between primary tumors and patient-matched metastases111. Therefore, evaluating frequently altered biomarkers in primary and metastatic tumors should be a priority before deciding on the immune checkpoint treatment to use in a particular BrM patient. The TME landscape112, genomic alterations113,114, and specific mutations involved in resistance to immune checkpoint therapies115 must also be taken into consideration to maximize therapeutic benefit from ICIs in the setting of BrM. Promising results seen in CAR T cell-based treatments of patients with hematologic cancers, as well as early clinical development in treating brain tumors, pave the way for more effective and precise immunotherapy.

In the final analysis, personalization is a true clinical need. New holistic and innovative approaches for BrM treatment using bispecific monoclonal antibodies and combination therapy may provide better response rates and survival benefits than those achieved by traditional single treatment strategies. Methods in the preclinical phase of the therapeutic pipeline such as targeting the promotion of M1 properties in macrophages within the TME and the recruitment of tumor-infiltrating CD8+ T cells, may also prove potent once translated to the clinic116. Meanwhile, emerging insights from the elegant immune landscape studies reviewed here are bound to impact our ongoing effort to convert BrMs from immunosuppressed “cold” to immunologically “hot” environments55.