Brain metastasis formation includes a highly selective, multistep process where circulating tumor cells (CTCs) must withstand hematogenous dissemination as they leave the primary tumor and enter blood circulation, arrest in the cerebral microvasculature, and traverse the BBB (22, 23). Following intravasation from the primary tumor, CTCs become sequestered in brain capillaries, and here the endothelial interactions and mechanical constraints determine their initial lodgment (22). For successful extravasation, tumor cells must engage and disrupt the specialized BBB architecture, which is filled with endothelial cells connected via tight junctions, in addition to pericytes, astrocytic end-feet, and basement membrane, that all collectively allows for CNS homeostasis, as discussed later in this section (24). Tumor-derived proteases, inflammatory cytokines, and exosomes can destabilize endothelial junctions, which reshapes the perivascular niche and aids in creating a heterogeneous blood-tumor barrier (BTB) that differs in permeability across metastases but is rather leaky in nature (25). The variability and disruption of the BBB is now better understood, with some BrMs maintaining an intact barrier while others exhibit substantial leakage. This notable heterogeneity impacts early BrM colonization, immune infiltration, and therapeutic penetration and potential (25, 26).

Upon entering the brain parenchyma, metastatic cells often co-opt existing vasculature, such as vessels and endothelial-astrocytic signals, to support initial survival and aid in motility (27, 28). Reactive astrocytes initially produce cytotoxic responses but transition quickly to a tumor-suppressive phenotype through cytokine secretion, like interleukin-6 (IL-6), interleukin-10 (IL-10) and transforming growth factor-beta (TGF-β). This promotes immune evasion and micrometastatic outgrowth (29–31). Microglia and infiltrating macrophages shape a permissive niche by releasing immunosuppressive mediators and remodeling extracellular matrix (ECM) structures (29).

Emerging evidence suggests that B cell dynamics further modulate these latter metastatic stages. In BrM particularly, B cells display functional heterogeneity. This is seen in variable functions including antigen-presenting and antibody-producing subsets with antitumor capacity, in addition to regulatory B cells (Bregs) which amplify IL-10-driven immunosuppression and inhibit T cell activity (21, 22). B cells also notably organize into tertiary lymphoid structures, which variably promote immune activation, or support metastatic progression depending on their respective cellular composition and cytokine profile (32, 33). Antibody-secreting B cells may also influence BrM growth by modulating complement pathways such as Fc-receptor signaling and myeloid polarization. Overall, the multitude of B cell functions, duality of responses, and mediation of antibody activation and presentation, cytokine signaling, and role in cross talk of processes together with blood vessels of the brain and support cells either facilitate the progression of small clustering to overt brain metastases, or conversely, promote anti-tumor immunity (23, 30, 33).

Circulating B cells, which tend to be largely naïve and classical memory subsets, differ functionally and phenotypically from tumor-infiltrating B cells, which are recognized to undergo local activation, clonal expansion, and differentiation within the TME (34, 35). Naïve (have not encountered a specific antigen) and memory cells (long-lived, antigen-specific) have distinct activation thresholds and antigen-presenting capacities that allow memory cells a quicker transition to antibody responses upon antigen re-exposure (36, 37). In solid tumors, and including BrM, infiltrating B cells tend to adopt class-switched memory or plasmablast/plasma cell phenotypes and can organize into TLS-like structures that perform local antigen presentation and promote vital T cell priming (34, 38, 39). Functionally opposed to this are Bregs, as mentioned above, which produce IL-10, TGF-β, or express PD-L1 and hinder effector T cells and dendritic cell functions (21, 22). Tumor-derived factors and stromal cues (i.e., IL-6, prostaglandins, CNS-specific astrocyte or microglial signaling) can promote Breg differentiation or expansion, thereby favoring local immunosuppression (40, 41). A combination of clinical and mechanistic evidence has suggested that a balance among these B cell subpopulations determines if B cells support anti-tumor immunity through antigen presentation and antibody production, or whether they impede anti-tumor immunity via Breg-mediated suppression. Thus, the phenotypic characterization of B cells is essential for potential prognostic and applications of B cell therapeutics (42).

In comparing BrM formation arising from different primary tumor types, it is important to recognize the tumor-origin-dependent immune and stromal programs (astrocyte-mediated processes in this case) that share similar properties while also diverging from the primary malignancy (43–45). Multiple cohorts and mechanistic studies have shown that lung- and melanoma-derived BrM tend to retain higher lymphoid activity, such as B and T cells and TLS signatures, in addition to immune-checkpoint pathway expression, whereas breast cancer-derived BrM are often more myeloid and neutrophil dominated with noted lymphoid scarcity (43, 45, 46). These patterns aid in explaining the varying responsiveness to immunotherapies across primary cancers (see clinical series and meta-analyses; JCO ASCO abstract) (44, 45). Recent immunoprofiling and spatial transcriptomic analyses have revealed distinct immune-stromal niches and TLS-associated programs enriched in lung and melanoma BrM, but downregulated in breast cancer-derived BrM (46, 47). From a mechanistic perspective, tumor cells from lung and breast primaries exploit CNS-specific stromal interactions, such as astrocyte-mediated STAT3 and survival signaling (e.g., BCL2L1. TWIST1, GSTA5 as discussed further in section 2.2.1), in order to promote colonization and therapy resistance within the brain microenvironment (43, 48). Overall, this highlights the primary-origin influence and brain-adaptation signaling that shape BrM biology and formation. Thus, therapeutic strategies should be CNS-specific and tailored to the tissue of origin to overcome shared stromal limitations.

The concept of CNS immune privilege emerged from experimental pursuits in the mid-20^th century (22). These were transplantation studies that showed grafts survived longer in the brain compared to peripheral sites, suggesting that the brain was immunologically unique (49). This apparent tolerogenic nature is conditional, because it remains vulnerable to activated peripheral immunity, where effector cells can further infiltrate the CNS. The immune privilege of the CNS arises from and is associated with structural and functional barriers. One distinguishing feature includes the BBB, as mentioned previously (50). Specifically, the BBB is formed by endothelial cells lining the cerebral microvasculature, interconnected by tight junctions, and supported by pericytes and astrocytic end-feet (51). Within the neurovascular unit, astrocytes exhibit a dual role which entails the initial release of neuroinflammatory mediators to restrict metastatic growth, but later shift toward a tumor-supportive phenotype by secreting growth factors and immunosuppressive cytokines, including TGF-β and IL-10 (52, 53). Through these cellular interactions, the BBB functions as an interface that regulates trafficking between the bloodstream and brain parenchyma on a molecular and cellular level. For this reason, the “barrier” presents a dual challenge in the BrM context. First, the BBB rigidly controls molecular agents and cellular immunotherapies, especially the delivery of antibodies or other immune modulators. In addition, the BBB limits immune cell infiltration, which leads to the development of an immune-privileged site where tumor cells can evade immune surveillance. The permeability of the BBB appears selectively variable (tumor, region, and molecule-specific) and heterogeneous. Some regions are impermeable, while others allow for selective penetration of therapeutics. Unlike glioblastoma (GBM), where neovascularization produces malformed and leaky vessels, the extent of neovascularization in BrM is less defined and may vary by tumor origin type and model system. This variability in BBB integrity fosters a spatial heterogeneity in immune cell infiltration and functioning throughout BrMs and presents complications for the efficacy and delivery of treatments. Notably, it appears that effects of brain metastasis on the leakiness of the BBB is cell line-dependent in preclinical models, where BBB can remain intact or are in a disrupted (50, 54, 55).

In the BrM context, the overall neurovascular unit interactions are significantly altered. Disruptions within the neurovascular interface are of concern because perturbations may promote tumor cell entry and reshaping of the respective microenvironment that could contribute to metastatic growth (56).

The brain’s immune landscape is distinguished by its dependence on resident immune cells and selective restriction of peripheral immune entry. Microglia and astrocytes constitute the main immune sentinels, although their activated states substantially differ compared to those of peripheral macrophages (57). In the periphery, macrophages tend to polarize along a spectrum between a pro-inflammatory state (M1), or a more anti-inflammatory, tissue-repairing state (M2). In CNS malignancies and BrM, microglia commonly induce a pro-tumor and anti-inflammatory, M2-like response, which hinders cytotoxic T cell recruitment and effector function (58). Responsible for this hindrance are the activated microglia and astrocytes which secrete immunosuppressive mediators such as IL-10 and TGF-β1. Unlike peripheral macrophages, CNS-resident myeloid cells are tightly associated with the neurovascular unit, where they aid in local inflammation regulation while also enforcing immune privilege by dampening effector responses.

Among the secreted mediators, TGF-β1 has emerged as a central orchestrator of immunosuppression in brain tumors. TGF-β1 is a modulator of proliferation, ECM deposition, and immune balance through canonical and non-canonical pathways (59, 60). It plays a context-dependent role in BrM, as it can facilitate a pro- or anti-tumor response. A notable role of TGF-β that amplifies the migratory and invasive tendencies of cancer is also suggested with it being an inducer of the epithelial-mesenchymal transition (EMT), where it suppresses immune cells in the brain, allowing for tumor cells to cross the BBB and enable metastatic progression. Its effects are context dependent because as it inhibits effector T cell proliferation and sustains the function of Tregs, it can simultaneously recruit macrophages, and in combination with IL-6, promote Th17 differentiation (59–61). As a result, this produces considerable amounts of IL-17 and maintains acute inflammation. In gliomas (primary glial-derived brain tumors), malignant cells exploit TGF-β1 in order to drive mesenchymal transition, invasion, and stem-like self-renewal. In BrM, astrocyte-derived TGF-β1 suppresses cytotoxic CD8⁺ T cell function and supports metastatic growth (62). Importantly, TGF-β1 also shapes B cell behavior within the BrM microenvironment (63, 64). As preliminarily suggested in recent work, spatial transcriptomic analyses in GBM show that myeloid cells are major sources of TGF-β1 activity within intratumoral B cell niches, suggesting that myeloid-B cell interactions are an important modulator of B cell suppression (64). This myeloid-driven TGF-β1-mediated suppression likely influences the regulatory B cell phenotype observed, which limits their capacity to stimulate T cells and form functional TLS (63, 64). Thus, as immunotherapies largely focus on targeting T cells, B cells in the TME are also impacted by TGF-β-dependent immunosuppression, which could halt APC function and hinder immune response (63, 65). Hou et al. recently demonstrated that dual blockade of αVβ8 integrin (activates TGF-β) and PD-1 restores B cell function, expands intratumoral B cells, and synergizes with PD-1 inhibition in order to eradicate nearly 60% of treated mice (64). This dual therapy was also shown to reduce B cell-mediated suppression of CD8⁺ T cell cytotoxicity and promote antitumor B cell activity (64). These findings underscore TGF-β as a key regulator of B cell proper functioning within CNS tumors and suggests that overcoming TGF-β-mediated B cell suppression could be critical in BrM, where there are similar myeloid-predominant niches and patterns of immune exclusion (64, 66). Clinical studies, involving patients with high-grade gliomas and brain metastases, further underscore the relevance of this mediator. For example, functional TGFB1 promoter polymorphisms correlate with greater circulating TGF-β1, adverse clinical features, and dismal survival trajectory in patients with brain malignancies (59). Overall, these findings position TGF-β1 as both a mechanistic facilitator of immune evasion and a possible prognostic predictor within the CNS tumor microenvironment, with its impact dependent on the context and possible tumor type.

BrM stem from peripheral cancers and overcome the structural and immunological barriers of the CNS in order to establish in the brain. The CNS imposes barriers to both immune trafficking and antigen flow, BBB and blood-cerebrospinal fluid (CSF), and has a distinct stromal milieu (astrocytes, microglia, perivascular macrophages) which shapes spatial patterns of immune exclusion that profoundly alter immune-tumor interactions (67, 68). By contrast, extracranial metastases arise in tissues with abundant lymphatic drainage and ongoing immune system influence, which allows for antigen presentation and T-cell priming. This means that BrM retain properties of their tissue of origin, they also rewire to survive within a more constrained immune landscape (69). This unique immune cell composition and spatial organization are two reasons why therapeutic responses in BrM diverge markedly from responses of corresponding extracranial tumors.

Brain metastases represent the product of convergent pressures imposed by the CNS microenvironment. In particular, melanoma brain metastases (MBM) highlight this principle, especially as representative in the work of Izar et al (70). MBM display pronounced chromosomal instability, micronuclei formation, and with notable neuronal-like differentiation programs (NGFR, NCAM1, NLGN3), which facilitate integration into neural circuits (70). MBM cells preferentially engage oxidative phosphorylation and PI3K/ERBB signaling, while also maintaining melanocytic lineage activity. This combination distinguishes this derivation of BrM from extracranial melanoma metastases (ECM), which instead favor EMT, AXL-drive invasive states (causes amplified invasiveness and drug resistance), and mTORC1 enrichment (70). Several of these features are also seen in other derivations of BrM, like breast and lung brain metastases. In these forms, there is also notable chromosomal instability, neuronal mimicry, and reliance on oxidative phosphorylation, which showcases the convergent metabolic and transcriptional alterations that support colonization of the neural niche. Across diverse BrM, shifting of the immune landscape toward monocyte-derived macrophages and the high expression of TOX (TOX⁺CD8⁺ T cells, associated with T cell dysfunction) appears to be a shared tendency (38). Compared to breast and lung BrM, melanoma BrM have robust lymphoid aggregates with evidence of B cell-to-plasma cell maturation (70, 71).

In contrast, GBM. the most common malignant primary brain tumor, diverges markedly from these metastatic developments and environments (72). Rather than adopting neuronal mimicry, GBM arises from neural precursor-derived stem-like populations and is sustained by transcriptional programs of plasticity (like SOX2, OLIG2, NANOG) that emphasize its proneural, mesenchymal, and classical subtypes (73, 74). Core drivers include PDGFRA/IDH1 alterations, NF1 loss, and EGFR amplification, which fuels Akt/STAT3 and NF-κB signaling pathways. Unlike BrM, GBM favors glycolysis, and thrives within the compact brain parenchyma, where extracellular space is limited. Glioblastoma stem cells (GSCs) occupy perivascular niches, where stromal and mesenchymal elements reinforce the “stem-like” state through activation of Wnt/Notch signaling and proteolytic remodeling of the extracellular matrix (which facilitates invasion) (75). Over time, selective stress exerted by treatments tend to promote a shift from proneural to mesenchymal states, a transition that fosters recurrence and resistance. Immunologically, GBM epitomizes the “cold” CNS tumor. This entails the TME being dominated by microglia and infiltrating macrophages, while adaptive immunity is constrained by T cell exclusion and immunosuppression driven by tumor-stromal cross-talk (76). This entrenched stromal and immune landscape juxtaposes that of BrM, which, while constrained by the CNS, retains greater inflammatory potential and lineage-specific immune engagement (Figure 1).

BrMs vary cellularly and molecularly from primary brain tumor developments. In BrM derived from different primary tumors such as lung, breast, and melanoma, there are notable differences in immune landscapes. BrM derived from lung cancer have higher expression of commonly mutated genes, particularly in genes that regulate immune checkpoint pathways, such as tumor necrosis factor receptor superfamily 9 (TNFRSF9), tumor necrosis factor receptor superfamily 4 (TNFRSF4), programmed cell death 1 ligand 2 (PDCDILG2), indoleamine 2,3-dioxygenase 1 (IDO1), inducible T cell costimulator (ICOS), and more (78). This suggests that lung cancer BrM could be increasingly sensitive or responsive to immunotherapy, compared to other BrM forms (78, 79). On the other hand, BrM stemming from melanoma tend to have a higher presence of CD4⁺ and CD8⁺ T cells, while breast cancer-derived BrM has a greater incidence of neutrophils and macrophages. This could suggest potential increased therapeutic resistance of breast cancer BrM juxtaposed to melanoma BrM (69, 78, 80). Lung cancer BrM has also been noted to have more T lymphocytes while melanoma has more B lymphocytes. Despite several differences between different microenvironments of BrMs derived from varying cancer forms, there remain similarities as well. As supported by in vivo experiments, both breast and lung cancer cells interact with astrocytes in the CNS in order to bolster the expression of survival genes, such as B cell lymphoma-2-like protein 1 (BCL2L1), twist family BHLH transcription factor 1 (TWIST1), and glutathione S-transferase alpha 5 (GSTA5) to acquire drug or therapy resistance (78, 81, 82). The variations in BrM microenvironments compared to those of primary brain malignancies provide important insights for treatment options, such as cancer-origin-dependent differences or mechanistic adaptation in the CNS, which necessitate therapeutic strategies tailored to BrM.

Foundational work in melanoma studies and other solid tumors has provided evidence that TLS and mature B cell states robustly correlate with response to immune checkpoint blockade (ICB). For instance, Helmink et al. demonstrated that activated and memory B cells within TLS predict ICB responses in melanoma (83), while Cabrita et al. showed that TLS density stratifies survival and therapeutic benefit (84). Similarly, Narvaez et al. suggested that TLS presence in breast cancer predicts improved survival and higher pathological complete response rates (85). These findings in conjunction with evidence suggesting B cells drive local antigen presentation, class switching, and CD4/CD8 T cell coordination contribute to a substantial framework for understanding how B cell immunity supports anti-tumor responses in extracranial sites (86).

Implications for BrM arise when these findings are juxtaposed with the niche constraints of the CNS. Although recent spatial and single-cell evidence has revealed that the presence of B cell aggregates and TLS-like structures in breast and lung-derived BrM, these intracranial TLS often exhibit incomplete organization, hindered germinal-center maturation, and altered chemokine expression (87). BrM studies have shown diminished levels of the B cell chemoattractant CXCL13, increased microglia and astrocyte-related suppression, and lymphocyte entry restrictions due to the BBB (84, 88). These features could limit the development of entirely functional TLS, thereby reducing the potential of B cells to harness ICB responses compared to melanoma or breast cancer in peripheral sites. Relevant findings from Mughal et al. have shown TLS signatures, which are observed in one-third of BrM and most frequent in lung cancer and melanoma metastases, mark tumors with strong B cell and T cell activation, including high immunoglobulin gene expression, mediated T cell recruitment, and lymphotoxin beta (LTß) upregulation (89). TLS-positive BrM also contains organized lymphoid aggregates, and specifically noted in lung cancer BrM patients, are strongly associated with prolonged survival, which underscores their relevance as a marker in BrM (89). Overall, the variance of cellular components of the tumor immune microenvironment in BrM is dependent on cancer type, and signify a key role of TLS formation for prognostic relevance and outcome in BrM patients (89).

Recent analyses from Nohira et al. suggest that B cell-rich immune niches could tangentially serve as prognostic markers in BrM (90). In a cohort of lung cancer BrM samples, TLS-like aggregates and associated B cell phenotypes, such as class switched memory states and germinal center-like activity, were detected in a subset of patients and were strongly associated with prolonged survival. TLS density, rather than the overall density of tumor-infiltrating lymphocytes (TILs), was the immune variable positively and independently correlated with patient outcome (90). This parallels the understanding in extracranial tumors, where TLS predict improved response to immunotherapy and promising prognosis, but applies these principles to the intracranial context where functional relevance of B cell immunity has remained relatively undefined. These patterns and findings suggest that B cell activation within the brain is shaped by neuroinflammatory cues along with tumor antigens, delineating the intracranial TLS biology from well-characterized TLS in extracranial malignancies (83–85, 90).

Altogether, these fundamental and emerging studies underscore that BrM exhibit a markedly distinct, CNS-influenced B cell landscape that contributes to or reshapes antigen presentation, antibody maturation, and response to ICB. Understanding these key differences between BrM and extracranial primary tumors suggests why BrM often respond poorly to current immunotherapies and presents a need for brain-specific strategies in order to induce functional TLS, enhance B cell recruitment and activation, and ultimately overcome neuroimmune suppression (89).

The discovery of immune checkpoint pathways has marked a shift in the cancer immunotherapy realm. Currently, the U.S. FDA has approved three major classes of ICIs including programmed cell death protein 1 (PD-1) inhibitors (Nivolumab, Pembrolizumab, Cemiplimab), programmed death-ligand 1 (PD-L1) inhibitors, and the cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) inhibitor (Ipilimumab) (102). These mechanisms dampen T cell activation, and the inhibitory receptors enable tumors to evade immune surveillance; so, blocking them restores and enhances anti-cancer immune (102). In the preclinical setting, previous studies utilized the CMT167 and LLC murine models of lung adenocarcinoma with high brain tropism, demonstrating that anti-PD-1/PD-L1 therapies could present a possible avenue for treatment of (103, 104). This anti-PD-1/PD-L1 therapy effectively reduced growth of primary orthotopic tumors, with up to 95% inhibition of CMT167 orthotopic tumor growth. There was also 30% inhibition of CMT167 subcutaneous tumors, and 35% inhibition of LLC orthotopic lung (103). With that being said, the brain tropic tendencies of cell lines could be harnessed to evaluate the efficacy in the BrM. In a tangential study, combination where ICIs, like anti-PD-1/PD-L1, are utilized in conjunction with an intratumoral injection of immune-activating agent (such as oncolytic viruses, and agonists), yielded synergistic anti-tumor effects, most likely caused by increased T cell infiltration and reduced (104). This was an attempt to convert “cold” tumors into “hot” immunologically responsive tumors. Mechanistically, these therapies re-energized exhausted immune cells within the brain TME, helping to facilitate cytotoxic (104, 105).

ICIs have also revolutionized treatment in the clinical setting. This is particularly seen in NSCLC patients afflicted with general metastases, which could show promise for BrM patients. For instance, the KEYNOTE-189 phase 3 trial probed the use of pembrolizumab, a PD-1 inhibitor, in combination with a platinum-based chemotherapy as a first- line treatment (106). There were improvements in overall survival (OS) and progression-free survival (PFS) over chemotherapy alone. CNS response rates were approximately 29.7% in PD-L1-positive patients. Many patients were able to avoid radiation treatment to the brain, and greater than 33% of the patients survived at least two years after the start of the clinical trial (106, 107). Additionally, the CheckMate 227 trial evaluated nivolumab (anti-PD-1) in combination with ipilimumab (anti-CTLA-4) in advanced NSCLC, including those with baseline treated brain metastases (108, 109). At a minimum follow-up of 5 years, BrM patients who received the combination ICI therapy (chemotherapy + ICI) had a median overall survival of 17.4 months compared with 13.7 months for solely chemotherapy (hazard ratio = 0.63), and demonstrated higher 5-year systemic (12% vs. 0%) and intracranial progression-free survival rates (16% vs 6%), which indicates improved long-term systemic and intracranial disease control (109).

The basis of TIL therapy involves isolating T cells from patient tumors, expanding them ex vivo, and reinfusing them to target malignancies. Its efficacy was first demonstrated in foundational melanoma trials, contributing to the development of adoptive cell therapy for solid tumors. Approval of lifileucel (Amtagvi) in February 2024 marked the first FDA-approved tumor-derived T-cell therapy for advanced melanoma, paving the way for its evaluation in preliminary contexts like BrM (110, 111). In the pivotal clinical study, phase 2 of C-144–01 trial published in 2022, lifileucel achieved systemic disease control, with a 31.4% objective response rate (ORR), including eight complete and 40 partial responses, with a median response duration exceeding three years; approximately one-third of responders remained progression-free at five years (112, 113). Although hematologic toxicity was experienced, responses to lifileucel were durable in a heavily pretreated population, where 81.7% of the trial population were pretreated with anti-PD-1 and anti-CTLA-4 agents (110).

Early-phase trials suggest that TIL therapy may also have activity in patients with CNS involvement. In the adoptive TIL therapy trial NCT02360579, lifileucel produced encouraging intracranial responses, some BrM cases, with minimal toxicity in metastatic melanoma (114). In addition, NCT03083873 was a phase 2 C-145–03 trial that evaluated autologous TIL therapy (LN-145), specifically in patients with recurrent or metastatic head and neck squamous cell carcinoma (HNSCC) (115, 116). Patients had surgical tumor resection to generate LN-145, followed by non-myeloablative lymphodepletion (NMA-LD) and IL-2 administration (115). Among the 52 patients treated, the ORR was 11%, the disease control rate was 76%, median OS was 9.5 months, and some responses, regarding systemic disease control, persisted beyond 23 months despite the heavily pretreated cohort (115). Collectively, these findings indicate that TIL therapy can harness durable antitumor activity systemically, and may retain functions within the CNS, supporting a preliminary relevance for BrM, although further intracranial studies remain limited compared to those of extracranial disease trials.

A recurring obstacle entails that of heterogeneity, inter- and intralesional, across BrM development. While the aforementioned targets, including PD-L1, EGFR/EGFRvIII, EpCAM, DLL3, and CD133, have shown relevance and promise in preclinical or clinical studies, the expression of these targets is often unfortunately transient. Tumors often “outsmart” the targeted immunotherapy. The antigen variability could be responsible for the development of adaptive resistance, essentially the tumor’s ability to adapt dynamically to the immune attack. Surviving tumor subclones are those that do not express the antigen, lack the therapeutic target, and expand or repopulate under immune pressure. For instance, in the case of EGFRvIII-directed CAR-T therapy, the loss of antigen expression was a prominent mechanism of recurrence in GBM, which raised similar concerns for the BrM context (126, 150, 151).

As the CNS TME is characterized as immunosuppressive, it is shaped by the interplay of blood-brain-barrier (BBB) restrictions, reduced T cell infiltration, and increased expression of suppressive inhibitory molecules, like IDO1, TGF-β, and STAT3 (142, 146, 147). This landscape continues to pose a challenge for therapeutic efficacy. Previously, the CNS was seen as “immune-privileged” because of the restrictive blood-brain and blood-CSF barriers, which presents a wall to entry of peripheral immune cells. However, recent discoveries of CNS lymphatic pathways and border-associated immune niches, like meninges and choroid plexus, have challenged this ideology, highlighting that brain malignancies aren’t entirely sequestered from systemic immunity (152).

Within BrM, tumor infiltration bypasses and disrupts the BBB, allowing the recruitment of peripheral immune cells, but also signaling chronic inflammation. The resident CNS cells, like astrocytes, microglia, and neurons, interact dynamically with infiltrating myeloid or lymphoid cells in efforts to curate an immunosuppressive niche. For example, the reactive astrocytes and microglia tend to release immunomodulatory factors, including TGF-β, IL-10 or VEGF, that weaken cytotoxic T cell activity while also promoting angiogenesis and tumor growth. Tumor-related macrophages and recruited myeloid-derived suppressor cells essentially foster this inhibitory environment, essentially pushing towards T cell exhaustion and diminished persistence (76, 152, 153).

This immunosuppressive crosstalk not only protects the metastatic cells from immune-mediated destruction, but also fosters metabolic stress, reduces antigen presentation, and accelerates therapeutic resistance. Notably, while TILs are often correlated with a more optimistic prognosis and immunotherapy response, in brain metastases, myeloid-lineage immune cells are found more predominantly compared to tumor-filtrating cytotoxic T cells. Rather attacking the tumor, the myeloid cells tend to adopt a more pro-tumor and immunosuppressive function, which fosters metastatic progression and therapeutic resistance (152). Therefore, the brain metastatic niche isn’t simply a passive sanctuary, rather serving as an active contributor in suppressing effective anti-tumor immunity. This poses a central barrier to harnessing a durable response across CAR-T therapy, ICI, and vaccine-based approaches.

The sequestered nature of the CNS vasculature, particularly the BBB and BTB, remain significant hurdles for adequate systemic therapy delivery. While locoregional strategies, like intraventricular and intratumoral delivery, help to improve trafficking, these approaches can be rather invasive and pose a concern for feasibility if patients have inaccessible or multifocal metastases. For instance, this delivery dependence was observed with the intravenous infusion of EpCAM or EGFR-directed CAR-T cells which yielded negligible intratumoral infiltration in preclinical murine BrM models, while local delivery was effective but lacked durability with a single dose (79).

Immunotherapy responses in BrM tend to be transient, with tendencies of relapse because of immune evasion or inadequate immunological memory formation. Vaccine-based platforms like DCVax-L or PerCellVac3 induce robust CD4⁺ and CD8⁺ T cell responses, as detailed previously, although their durability in the CNS remains unclear, especially given the immunological “coldness” of the brain and associated malignancies. Similarly, BiTEs like DLL3-targeted tarlatamab display optimistic efficacy in SCLC with BrM, but responses are often short-lived without continued infusion. In the tarlatamab trial, the majority of the patient cohort was also heavily pre-treated, which could have played into the effects and nuances of the immunotherapy application (121).

Although mostly tolerable, immunotherapies can induce side effects and potential severe neurological toxicities within the BrM context. With the application of ICIs, there have been instances of exacerbated cerebral edema and increased necrosis risk when in combination with stereotactic radiosurgery (SRS). Seizures were an initial symptom in a striking 40% of melanoma brain metastatic (MBM) patients (154), which may also be aggravated by ICIs, resulting in the need for antiepileptic drug usage during some trials (155, 156). Symptomatic edema was also reported in the CheckMate-204 trial in particular, with incidence ranging up to 36% with combined ipilimumab and SRS in melanoma-derived brain metastases (157, 158). Baseline edema often does not impact anti-PD-1 response in melanoma or NSCLC patients (159), although symptomatic edema often requires ICI interruption, corticosteroids, and supplementary local therapy including surgery or radiation treatment (156). Corticosteroids are known to be widely used in BrM, and patients with primary brain tumors, for symptomatic relief of intracranial pressure or edema, with recommended doses starting at 4–8 mg/day of dexamethasone and potential escalation to ≥16 mg/day for optimal mass effect (160, 161). Baseline or high-dose systemic steroid use has been associated with unfortunately reduced ICI efficacy and lower PFS/OS in metastatic NSCLC and melanoma, as suggested by a meta-analysis in BrM patients with reported shorter systemic PFS (Hazard Ratio ~2.0) and worse OS (Hazard Ratio ~1.84) with steroid use (160, 161). Corticosteroids are also utilized in patients with GBM for edema control, and their effects are correlated to reducing vasogenic edema rather than exerting antitumor activity (162). Additionally, CAR-T and BiTE cell therapies can trigger and induce cytokine release syndrome (CRS). CRS is a cascade of toxic adverse events seen during infection or after antibody administration for therapeutic testing. It is a response associated with the high circulating concentrations of different pro-inflammatory cytokines, like stimulating factors, necrosis factors, and more (163, 164). There have been recent efforts to prevent and control CRS, although it is still an unmet clinical need.

With the plethora of studies analyzed above, BrM patients remain still underrepresented in clinical trials. Many pivotal studies exclude those with symptomatic and untreated CNS metastases, which poses a concern for widespread efficacy. Current evidence originates from trials with smaller cohorts, fewer randomized trials, post-hoc analyses, or from trials that are in the exploratory stage. This poses a major concern in generalizing findings for treatments tailored to patients afflicted with brain metastases.

B cell-based immunotherapies have the possibility to overcome several limitations of T cell-focused approaches for treating BrM. T cells in the CNS often exhibit notable functional exhaustion, limited antigen access, and hindered recruitment because of the BBB and immunosuppressive astrocyte-myeloid networks (177). However, B cells offer complementary functions such as local antigen presentation, intratumoral antibody production, cytokine polarization, and facilitation of TLS formation which could circumvent some of these recognized barriers and present therapeutic opportunity. Preclinical work with BVax and 4-1BBL+/CD40-activated B cells, as discussed previously, shows that ex vivo-B cells can drastically enhance CD8⁺ T cell priming, encourage cytotoxic responses, and shape suppressive niches (148, 149). These insights show promise for B cells as a possible synergistic partner capable of improving checkpoint efficacy or restoring activity in immune-restricted BrM. It is important to acknowledge that translational challenges remain, especially due to BrM being often in accessible, unresected, or biopsy-limited, which restricts high-resolution profiling of spatially organized B cell states, TLS formation, and Breg-influenced niches (178). Circulating B cell markers provide promise, but not the whole picture, of understanding the role of B cells within BrM and the CNS immune environment. Within BrM, infiltrating B cells, such as naïve, plasma, and memory B cells, in tumors are positively correlated with patient prognosis, and the overarching heterogeneity of subpopulations provides insight into the degree of anti-tumor response (42). The presence, density, and maturation of TLSs have also been demonstrated to be notable prognostic markers (42). Overall, this review denotes a vital therapeutic gap in BrM management and proposes leveraging B cells for treatment as a promising avenue to improve the prognosis of BrM-afflicted patients.