Melanoma leptomeningeal disease: Advances in diagnosis and emerging therapeutic strategies

Abstract

Melanoma leptomeningeal disease (LMD) is a rare and severe manifestation of advanced melanoma involving malignant infiltration of the leptomeninges and cerebrospinal fluid. Historically, survival was measured in weeks, but recent advances in diagnostics and therapy have begun to improve outcomes. This narrative review summarizes current evidence on the epidemiology, biology, diagnosis, and management of melanoma LMD, with emphasis on studies from the past five years. Emerging cerebrospinal fluid liquid biopsy methods, including circulating tumor cell and cell free DNA analysis, allow earlier detection, identification of actionable mutations, and real time monitoring of disease activity. Modern systemic therapies have extended median overall survival to approximately 8.4 months compared with historical outcomes of 2.9 months to 3.7 months. Immune checkpoint inhibitors and B-rapidly accelerated fibrosarcoma and mitogen-activated protein kinase targeted therapies both contribute to these improvements, and early studies of intrathecal immunotherapy show encouraging clinical activity. Additional innovations such as proton craniospinal irradiation and novel drug delivery strategies reflect an evolving treatment landscape. Continued progress will depend on dedicated clinical trials, broader inclusion of patients with LMD, and deeper understanding of the leptomeningeal immune microenvironment.

Key Words: Melanoma leptomeningeal disease; Leptomeningeal metastases; Cerebrospinal fluid diagnostics; Intrathecal immunotherapy; Central nervous system metastases

Core Tip: Melanoma leptomeningeal disease (LMD) is the most severe complication of advanced melanoma, characterized by very poor outcomes and a lack of robust evidence to inform management. This narrative review synthesizes recent advances in the understanding of LMD biology, highlights the clinical significance of evolving diagnostic strategies including neuroimaging and cerebrospinal fluid-based assays, and reviews emerging therapeutic approaches. The article contrasts LMD with parenchymal brain metastases, identifies ongoing clinical challenges, and notes key research priorities needed to improve outcomes for this historically understudied condition.

INTRODUCTION

Melanoma leptomeningeal disease (LMD) is defined as the infiltration of malignant melanoma cells into the leptomeninges (arachnoid and pia mater) and the cerebrospinal fluid (CSF) compartment[1]. This process occurs through hematogenous dissemination or direct extension from parenchymal brain metastases, leading to diffuse or multifocal seeding of tumor cells throughout the neuraxis[1,2]. Clinically, this results in rapid neurological decline and is associated with a very poor prognosis, with median overall survival measured in weeks to only a few months, even with aggressive management[3].

Diagnosis is challenging, as symptoms are often nonspecific, and confirmation typically requires both magnetic resonance imaging (MRI) and CSF cytology[4-6]. The unique immune-suppressive microenvironment of the leptomeninges, characterized by enrichment of myeloid-derived suppressor cells (MDSCs) and exhausted T-cells, further contributes to therapeutic resistance[1,7]. Traditional therapeutic approaches such as radiation therapy and chemotherapy have been frequently employed, but their impact on survival is minimal[4,6]. Systemic therapies, including immune checkpoint inhibitors (ICIs) and B-rapidly accelerated fibrosarcoma (BRAF)/mitogen-activated protein kinase (MEK) inhibitors in BRAF-mutant melanoma, have modestly extended survival compared to historical controls, though outcomes remain poor[8-11].

Recent years have seen a surge of investigational strategies aimed at improving outcomes in this population. Intrathecal (IT) immunotherapy with programmed death 1 (PD-1) inhibitors (nivolumab, pembrolizumab) has demonstrated feasibility, safety, and preliminary signals of clinical benefit[12,13]. Novel modalities such as proton craniospinal irradiation (pCSI) and molecularly guided therapies are also under investigation[14,15]. Overall, emerging strategies including CSF liquid biopsy-guided targeted agents, cellular therapies, and systemic-IT combinations reflect an active area of investigation that continues to evolve.

This article synthesizes current knowledge of melanoma LMD, with a focus on evolving diagnostics, novel therapeutic approaches, and the unique biology that drives resistance. By outlining both present challenges and future opportunities, we aim to provide a framework for advancing clinical care and research in this devastating condition.

EPIDEMIOLOGY AND CLINICAL PRESENTATION

Epidemiology

LMD is a recognized but relatively uncommon complication of systemic malignancy, with an overall incidence estimated at 5%-15% across cancer types[9,16,17]. Although some reviews have reported a slightly lower range of 3%-5%, the broader consensus in recent literature supports the higher 5%-15% figure as the most accurate estimate[18]. Within melanoma specifically, LMD develops in approximately 10%-15% of patients with advanced disease, underscoring its clinical significance despite its relative rarity[3].

The demographic profile of melanoma LMD shows that it largely affects adults in mid to later life. Retrospective cohorts report median ages of 51 years and 58 years, placing most patients in their fifties[8,19]. This aligns with advanced melanoma overall, though LMD often emerges relatively early within the metastatic course.

From a molecular standpoint, melanoma patients who develop central nervous system (CNS) involvement frequently harbor BRAF V600 mutations, with greater than 50% of cases demonstrating this alteration[8,20]. This finding is clinically relevant given the therapeutic implications of BRAF/MEK inhibition, though the effectiveness of targeted therapy in the leptomeningeal compartment remains an area of ongoing investigation.

Clinically, melanoma LMD is usually symptomatic at presentation. Data shows that 85% of patients reported neurologic symptoms, with 69% showing concurrent parenchymal metastases and 31% presenting with isolated LMD[8,10]. This distribution illustrates the diffuse pattern of leptomeningeal involvement. These patterns of presentation have important implications for both diagnosis and prognosis, as coexisting brain metastases often complicate treatment strategies and contribute to the overall poor outcomes associated with melanoma LMD.

Clinical presentation

The clinical presentation of melanoma LMD is highly variable and often multifocal, reflecting the diffuse involvement of the CNS. Symptoms typically develop in a subacute and progressive manner, and may initially be subtle and nonspecific, contributing to delays in diagnosis. The most common presenting features include headache, nausea, vomiting, diplopia, and limb weakness, at least one of which are present in approximately 85% of cases at diagnosis[1,8,21]. Other frequently reported manifestations are altered mental status, visual disturbances, sensory deficits, and gait instability[8,22].

Cranial nerve palsies are a particularly notable component of the clinical spectrum, with the oculomotor (III), abducens (VI), and facial (VII) nerves most often affected. These deficits commonly manifest as diplopia, ptosis, ophthalmoplegia, and facial weakness, significantly impairing quality of life[23-25]. Signs of increased intracranial pressure are also prevalent and clinically significant, arising from disruption of CSF flow secondary to leptomeningeal tumor infiltration. Papilledema observed on ophthalmologic examination may even serve as the first clue to LMD in the absence of other neurological deficits[1,26,27]. Prompt recognition of these findings is critical as they are associated with rapid disease progression and poor prognosis.

Seizures may also occur in this setting. Recent reports suggest a prevalence of approximately 22% among patients with LMD from solid tumors overall, though this rate does not appear to be disproportionately higher in melanoma compared to other malignancies[28]. Importantly, the presence of parenchymal brain metastases represents a more significant risk factor for seizures than leptomeningeal involvement itself[28].

Finally, case reports highlight that atypical or isolated findings, such as hemorrhagic CSF, may be the first manifestations of melanoma LMD, which further underscores the diagnostic challenges[29]. Collectively, these patterns illustrate the variability of melanoma LMD at presentation and highlight the need for close clinical attention to ensure timely diagnosis.

Prognosis

Melanoma LMD carries an extremely poor prognosis, with median overall survival typically measured in weeks to a few months. Across contemporary cohort studies, median survival generally ranges from 2.9 months to 8.4 months despite advances in systemic therapy[8,10,11]. Population-level data suggest that modern treatments including ICIs and BRAF/MEK-targeted therapy have modestly extended survival relative to historical benchmarks. Despite this, long-term survival remains rare. Several studies demonstrate that trial of any active treatment including systemic therapy, radiotherapy, or IT immunotherapy is associated with improved survival compared with no treatment, although the magnitude remains limited.

Prognostic factors are relatively consistent across cohorts. Poor performance status, neurological symptoms at diagnosis, elevated LDH, and signs of widespread or progressive disease are consistently associated with shorter survival. In contrast, patients who maintain functional status well enough to receive multimodal therapy tend to live somewhat longer. Table 1 summarizes major cohort studies evaluating outcomes in melanoma LMD and highlights the narrow overall survival range and prognostic patterns observed across institutions.

Table 1 Summary of survival outcomes in melanoma leptomeningeal disease cohorts.

Ref.	n	Median OS	Treatment groups (summary)	Prognostic factors
Pedersen et al[11], 2025	67	8.4 months	ICI or BRAF/MEK improved OS; no OS difference between regimens	LMD independently worsens OS; ECOG ≥ 2, high LDH, male sex
Nyman et al[36], 2023	86	3.7 weeks untreated; 10.8 weeks treated	Systemic, intrathecal, or radiation therapy improved OS	Prior therapy; systemic treatment exposure
Chorti et al[8], 2021	52	2.9 months; 3.7 months with therapy	Modest OS benefit from targeted therapy or ICI	Elevated LDH; ECOG ≥ 2
Ferguson et al[19], 2019	178	3.5 months	Targeted and intrathecal therapy prolonged OS	Higher ECOG; neurologic symptoms; systemic disease protective
Tétu et al[21], 2020	29	5.1 months; 7.1 months with systemic + RT	Systemic + radiotherapy improved OS	High LDH; neurologic symptoms
Geukes Foppen et al[10], 2016	39	6.9 weeks; 2.9 untreated; 16.9 treated	Targeted therapy and immunotherapy improved OS	High LDH; high S100B
Harstad et al[121], 2008	110	10 weeks	Intrathecal chemotherapy associated with longer OS	High LDH; cytology not prognostic

OS: Overall survival; ICI: Immune checkpoint inhibitor; BRAF: B-rapidly accelerated fibrosarcoma; MEK: Mitogen-activated protein kinase; LMD: Leptomeningeal disease; ECOG: Eastern Cooperative Oncology Group; LDH: Lactate dehydrogenase; RT: Radiation therapy.

Comparison of LMD and parenchymal brain metastases

LMD results from dissemination of malignant melanoma cells into the leptomeninges and CSF compartment, resulting in diffuse or multifocal tumor infiltration of the neuraxis. The immune microenvironment of LMD is uniquely suppressive, characterized by MDSCs and exhausted T-cell populations, while cytotoxic CD8+ T cells are relatively scarce[30]. In contrast, melanoma brain metastases (MBM) represent discrete tumor deposits within the brain parenchyma, most commonly arising at the gray-white matter junction. These lesions often manifest with focal neurological deficits, seizures, or symptoms related to mass effect, and the immune milieu is comparatively less suppressive, making them more responsive to both local therapies [surgical resection, stereotactic radiosurgery (SRS)] and systemic immune-based approaches[31,32].

From an epidemiologic standpoint, MBM are far more common, occurring in 30%-50% of patients with advanced melanoma, whereas LMD develops in approximately 5%-15% of cases and typically represents a late-stage manifestation of disease[3,33,34].

Diagnostic approaches also differ between the two entities. LMD is most reliably diagnosed by identifying malignant cells in the CSF via cytology, which remains the gold standard, although gadolinium-enhanced MRI is used to detect characteristic leptomeningeal enhancement[4]. In contrast, MBM are primarily diagnosed by neuroimaging, with MRI revealing discrete contrast-enhancing parenchymal lesions, frequently located at the gray-white matter junction[31,35]. CSF analysis is not routinely required in MBM unless there is suspicion for concomitant leptomeningeal involvement[31].

Prognosis highlights perhaps the most striking difference between these two forms of CNS involvement. Survival in melanoma LMD remains short, typically under six months even with treatment[8,19,36]. In contrast, outcomes for MBM have improved dramatically in the era of ICIs and BRAF/MEK-targeted therapy, with modern cohorts reporting median overall survival ranging from 8.9 months to 16.6 months and a subset of patients achieving durable long-term survival[31,37,38]. Importantly, the development of LMD in patients with MBM is an independent predictor of significantly worse outcomes[31].

Together, these distinctions underscore that while both LMD and MBM represent severe manifestations of advanced melanoma, they differ in frequency, diagnostic approach, therapeutic responsiveness, and prognosis, with LMD remaining the prognostically worst presentation.

PATHOPHYSIOLOGY AND DISEASE BIOLOGY

Pathophysiology of melanoma cell spread to leptomeninges

Melanoma cells can reach the leptomeninges through several distinct pathways (Figure 1). Hematogenous dissemination allows circulating tumor cells (CTCs) to adhere to and transmigrate across the blood-brain barrier (BBB) and blood-CSF barrier, a process facilitated by disruption of endothelial tight junctions and the release of proteolytic enzymes such as seprase, which degrade extracellular matrix and junctional complexes to enable paracellular migration[39]. Direct extension from dural or parenchymal metastases from lesions near CSF spaces is another common route, while iatrogenic seeding during surgical procedures that breach the CSF compartment represents a less frequent mechanism[16]. Rarely, primary leptomeningeal melanoma may arise from resident melanocytes within the leptomeninges themselves[16].

Figure 1 Pathophysiology of melanoma leptomeningeal disease. Created with BioRender.com. Melanoma cells disseminate from the primary tumor via hematogenous spread and traverse the central nervous system vascular-cerebrospinal fluid (CSF) interface to access the leptomeningeal space. Tumor cells spread diffusely along the pia and arachnoid mater within the CSF. The CSF microenvironment is marked by low immune surveillance, enriched transforming growth factor β signaling, scarce cytotoxic CD8+ T cells, abundant myeloid-derived suppressor cells, and exhausted CD4+ T cells, promoting tumor persistence and therapeutic resistance. CSF: Cerebrospinal fluid; CNS: Central nervous system; CTCs: Circulating tumor cells; MDSCs: Myeloid-derived suppressor cells; TGF: Transforming growth factor; ICIs: Immune checkpoint inhibitors.

At the molecular level, leptomeningeal dissemination reflects a multistep process driven by various genomic alterations that promote tumor survival, invasion, and immune evasion. Activating mutations in mitogenic driver genes such as BRAF or NRAS initiate constitutive MAPK pathway signaling, while loss of cell-cycle regulators like cyclin-dependent kinase inhibitor 2A enables escape from oncogene-induced senescence[40,41]. Alterations including TP53 dysfunction and telomerase reverse transcriptase promoter mutations facilitate resistance to apoptosis[40,42]. Together, these events support sustained proliferation and metastatic competence. Crosstalk between the MAPK and phosphoinositide 3-kinases (PI3K)/protein kinase B (AKT)/mammalian target of rapamycin pathways further enhances tumor survival under metabolic and immune stress, conditions that are particularly relevant within the leptomeningeal compartment[40,43]. Consistent with this biology, melanoma cells in the CSF typically retain expression of melanocytic lineage markers including S100 calcium-binding protein, SRY-Box transcription factor 10, MART1, and human melanoma black-45, supporting both diagnostic confirmation and continuity between systemic and LMD[44,45].

Once within the CNS, melanoma cells show a predilection for regions adjacent to CSF spaces. High-resolution MRI studies demonstrate that more than 90% of small intracranial melanoma metastases occur in close proximity to the leptomeninges, supporting the concept that this compartment provides a permissive microenvironment for metastatic seeding[46]. Anatomical factors such as dural-based or intraventricular metastases further increase the likelihood of leptomeningeal dissemination[16].

Survival and proliferation within the leptomeningeal space pose unique biological challenges given that CSF is composed of 99% water with low concentrations of nutrients, proteins, and cytokines relative to plasma[47,48]. To thrive in this nutrient-poor environment, melanoma cells rely on adaptation to local metabolic and signaling cues. Proteomic analyses of CSF from patients with LMD have identified enrichment of pathways promoting innate immunity, cell adhesion, platelet activation, insulin-like growth factor 1, glycogen synthase kinase 3 beta, and Notch signaling, with marked upregulation of complement proteins including C2, C3, C5, and C9[49]. Complement component C3 has been implicated in disrupting the blood-CSF barrier via activation of C3a receptors on choroid plexus epithelial cells, facilitating tumor cell invasion and creating a supportive microenvironment for growth[48].

The immune microenvironment of the leptomeninges further contributes to tumor persistence and therapeutic resistance. Single-cell RNA sequencing studies show that the leptomeninges are dominated by dysfunctional immune populations, including exhausted CD4+ T cells, scarce CD8+ T cells, and expanded MDSCs and macrophages with signal transducers and activators of transcription 3-associated signaling[7,30,49]. Similar immunosuppressive features have been described in LMD from other solid tumors, but they appear more prominent in melanoma and are associated with worse clinical outcomes[7,50,51]. Importantly, a small subset of patients with prolonged survival beyond three years show CSF immune profiles that more closely resemble non-LMD controls, with higher CD4+ and CD8+ T-cell levels and increased dendritic cells, B cells, and plasma cells[30].

Finally, soluble mediators in the CSF play an additional role in shaping therapeutic resistance. CSF from patients with aggressive LMD has been shown to reduce melanoma cell sensitivity to MAPK-targeted therapy by promoting tumor survival signaling, in part through increased transforming growth factor (TGF) β1 activity and downstream AKT phosphorylation and integrin pathways[49]. Elevated CSF TGFβ1 Levels have been correlated with worse outcomes, suggesting that modulation of this pathway could represent a future therapeutic strategy[49,52,53].

Together, these findings indicate that leptomeningeal colonization is not driven solely by anatomic access to the CSF. Instead, melanoma cells must also adapt to a nutrient-poor yet immunosuppressive microenvironment that supports tumor survival, immune escape, and resistance to therapy. Primary meningeal melanocytic tumors, including diffuse meningeal melanocytosis and melanomatosis driven by GNAQ or GNA11 mutations, represent biologically distinct entities from metastatic melanoma LMD and are beyond the scope of this article.

Barrier to therapy

One of the major challenges in treating melanoma LMD is the limited penetration of therapeutic agents into the CNS. The CNS is protected by two specialized barriers: The BBB and the blood-CSF barrier (BCSFB). The BBB is formed by tightly joined endothelial cells reinforced by pericytes and astrocytic end-feet, creating a structure that minimizes passive diffusion and actively effluxes many agents through transporters. The BCSFB, located at the choroid plexus, is composed of fenestrated capillaries covered by a layer of epithelial cells joined by tight junctions which further regulate molecular entry into the CSF[54,55]. These barriers restrict the access of most systemic agents including small-molecule targeted therapies and large monoclonal antibodies into the leptomeningeal compartment, resulting in subtherapeutic drug concentrations[56].

Although focal disruption of these barriers can occur in areas of bulky parenchymal metastases, the diffuse and microscopic nature of LMD means that many tumor cells remain protected in regions of intact barrier integrity, especially within the subarachnoid space[54]. This distinction helps explain why systemic therapies effective against extracranial and parenchymal brain disease frequently fail to demonstrate the same efficacy in LMD. For example, BRAF and MEK inhibitors show limited CSF penetrance, and antibodies such as anti-PD-1 and anti-cytotoxic T-lymphocyte-associated protein 4 have poor bioavailability due to their molecular size and inability to cross intact barriers[55,56].

Attempts to circumvent these obstacles with IT administration provide direct access to the CSF but come with unique limitations. Chemotherapy agents delivered intrathecally distribute unevenly along the craniospinal axis and may accumulate in areas of impaired CSF flow, reducing efficacy and increasing local toxicity. Moreover, melanoma has shown relative resistance to IT chemotherapy compared with hematologic malignancies which has limited the clinical impact of this strategy[54]. Similarly, IT immunotherapy remains investigational, with early studies suggesting feasibility but underscoring the challenges of safely delivering biologics into this compartment.

CSF flow dynamics themselves further complicate drug delivery. CSF is produced primarily by the choroid plexus at a rate of approximately 300 mL/day and circulates through the ventricles into the subarachnoid space before draining into the venous system via arachnoid granulations[57,58]. However, CSF flow is not uniform. It is pulsatile and varies by region, and in LMD this flow is often altered by tumor infiltration or secondary hydrocephalus, resulting in uneven drug distribution and relative underexposure of tumor cells in certain compartments.

Altogether, these structural, physiological, and immunological features of the CNS create formidable barriers to therapy in melanoma LMD. Overcoming these challenges requires innovative strategies to enhance CNS penetration, optimize IT delivery, or exploit alternative transport mechanisms across the BBB and BCSFB.

DIAGNOSIS: EVOLVING APPROACHES

Traditional tools

CSF cytology remains the gold standard for diagnosing melanoma LMD. It involves microscopic examination of CSF obtained by lumbar puncture for malignant cells, but sensitivity is limited, approximately 75% for a single sample, with false negatives common early in the disease course or when tumor burden is low[17,59,60]. Diagnostic yield improves with repeated sampling, larger CSF volumes, and collection near the site of radiographic or symptomatic disease. Despite its specificity, cytology may remain negative in up to 10% of cases with strong clinical or radiographic evidence of LMD, necessitating adjunctive diagnostic modalities[4,17]. Malignant melanoma cells in the CSF are typically large and pleomorphic with prominent nucleoli, and may contain melanin pigment, although pigment is not consistently present. Ancillary techniques such as immunocytochemistry with markers including human melanoma black-45 and S100 calcium-binding protein can increase diagnostic specificity when cytomorphology is equivocal[61]. Importantly, CSF cytology is mandatory whenever leptomeningeal metastasis is suspected, even if standard CSF parameters such as cell count, protein, and glucose are normal[5,62].

MRI with gadolinium contrast is the primary imaging modality used to assess suspected LMD. MRI may show leptomeningeal enhancement, nodularity, or secondary findings such as hydrocephalus, but specificity is limited because similar enhancement can be seen after surgery, infection, or radiation, and imaging alone cannot reliably distinguish LMD from these entities[4,5]. Sensitivity is also imperfect; MRI can be negative in patients with cytologically confirmed LMD, and radiographic findings often do not correlate with CSF parameters or clinical symptoms. Comparative studies suggest gadolinium-enhanced MRI has a sensitivity of about 76% and a specificity of about 77% when benchmarked against CSF cytology[59]. Nonetheless, MRI is valuable in detecting disease even when cytology is negative and provides critical information on the anatomical extent of involvement to guide both diagnosis and therapy[63].

Diagnostic accuracy is further influenced by imaging technique. High-resolution, thin-slice, post-contrast T1-weighted MRI, particularly 3D sequences (T1-SPACE, T1-mVISTA, and 3D T1 fast field echo), provides higher sensitivity and better interobserver agreement than conventional 2D T1-weighted or FLAIR imaging[64,65]. Use of thin slices and minimal or no interslice gap further improves detection rates and is recommended for both initial diagnosis and accurate mapping of disease burden, which is critical for therapeutic planning[66]. The choice of contrast agent affects sensitivity, as gadobutrol is associated with improved detection of subtle meningeal metastases compared with gadopentetate dimeglumine[67]. High-dose gadolinium protocols (0.3 mmol/kg) can increase further sensitivity in select cases by revealing enhancement not visible at standard doses, though their routine use is limited by safety concerns and lack of prospective validation[68,69].

Spinal MRI is an important adjunct, as leptomeningeal involvement frequently extends along the spinal axis, sometimes in the absence of clinical symptoms. In one series, enhanced spinal MRI detected leptomeningeal metastases in 60% of patients with negative CSF cytology and in 49% of patients without spinal symptoms, emphasizing its value in comprehensive staging[63].

In summary, CSF cytology provides definitive diagnosis when malignant cells are identified, but its sensitivity is suboptimal and often requires repeat sampling. MRI complements cytology by offering anatomical localization and disease mapping, though it lacks specificity and can be negative in cytology-confirmed cases. Together, these two modalities remain the cornerstone of diagnosis.

Emerging tools

Cell-free DNA/circulating tumor DNA: Liquid biopsy of the CSF has rapidly become an important diagnostic and research tool in melanoma LMD. Among these approaches, analysis of cell-free DNA (cfDNA) and circulating tumor DNA (ctDNA) has shown particular promise for detecting disease and defining its molecular landscape.

Next-generation sequencing and droplet digital PCR of CSF-derived cfDNA can identify tumor-specific mutations such as BRAF and NRAS with markedly greater sensitivity than conventional cytology, and performance is not dependent on the proportion of tumor cells present in the sample[17,70,71]. In recent studies cfDNA sequencing achieved 100% sensitivity for detecting known driver mutations in patients with LMD, compared with 82% for cytology and 93% for CTC analysis[70]. Variant allele frequencies measured in CSF cfDNA are often higher than those detected in cellular DNA, suggesting preferential capture of tumor-derived DNA with relatively limited background from nonmalignant cells[70,72].

Comprehensive cfDNA profiling enables molecular characterization of LMD beyond simple detection. Sequencing can reveal actionable mutations, track emerging resistance mechanisms, and delineate tumor heterogeneity, providing insights that may guide targeted therapy selection[73,74]. Longitudinal analysis of cfDNA also correlates with disease trajectory, as mutant allele fractions typically decline in patients responding to therapy and rise again with relapse[74]. From a practical standpoint, cfDNA analysis offers additional advantages as extracted DNA remains intact for delayed analysis, in contrast to cytologic preparations, which are more vulnerable to degradation and handling variability[17].

Nonetheless, several limitations remain. False-positive results can occur in patients with parenchymal brain metastases adjacent to CSF spaces, as DNA from nearby lesions may be released into the fluid[17,70]. Furthermore, the clinical meaning of isolated cfDNA positivity, especially in the absence of radiographic or cytologic evidence of leptomeningeal involvement has yet to be fully established[72].

Overall, CSF cfDNA and ctDNA analysis represent a substantial advance over traditional cytology, combining high diagnostic sensitivity with the ability to profile tumor biology and therapeutic response. Continued refinement of these assays may ultimately allow for earlier detection of LMD, real-time monitoring of treatment efficacy, and more personalized therapeutic decision-making.

CTC analysis: CTC analysis in melanoma LMD is performed using immunoassays that target melanoma-specific antigens such as CD146 and high-molecular weight melanoma-associated antigen, while excluding hematopoietic cells (CD45-) to accurately distinguish malignant cells within the CSF[75,76]. The Cell Search platform and newer rare cell capture technologies have been adapted for CSF, enabling enrichment, visualization, and enumeration of melanoma CTCs. Unlike epithelial cancers, melanoma does not express EpCAM meaning these assays rely on melanoma-specific surface markers for cell identification[75,76].

The diagnostic performance of CTC analysis in LMD is high. Reported sensitivities range from 91% to 100% in solid tumors (including melanoma) with specificity approaching 97%-100%[76,77]. This level of accuracy substantially exceeds that of conventional CSF cytology, which typically achieves sensitivities of only 33%-82%[70,76]. Importantly, CTC detection can confirm LMD even in patients with negative MRI or cytology results, providing an additional diagnostic avenue for early disease detection[78].

Beyond diagnosis, CTC analysis offers several clinical applications. CTC enumeration provides a surrogate measure of tumor burden and has been consistently associated with prognosis, with higher CTC counts linked to shorter overall survival[78]. The method can also reduce the need for repeated lumbar punctures by enabling earlier and more accurate diagnosis of LMD[75,76]. Furthermore, isolated CTCs can be subjected to molecular profiling and ex vivo culture, enabling preclinical studies of signaling pathways and therapeutic vulnerabilities[79].

When compared with other CSF liquid biopsy modalities, such as cfDNA or ctDNA analysis, CTC detection provides complementary information. Although cfDNA/ctDNA analysis is generally more sensitive for mutation detection and independent of the presence of intact tumor cells, CTC analysis uniquely allows quantification of viable cells and direct morphological assessment[70,78,80]. Each approach therefore contributes distinct advantages. cfDNA excels in identifying driver mutations and tracking clonal evolution, whereas CTC analysis better reflects viable tumor burden and functional status.

Proteomic profiling of CSF: Proteomic profiling of CSF in melanoma LMD is an emerging approach that characterizes the unique biology of the leptomeningeal microenvironment and identifies protein signatures linked to disease progression, therapeutic resistance, and prognosis. Mass spectrometry studies of CSF in melanoma LMD show that patients with poor survival have increased expression of proteins involved in innate immune signaling, tissue proteolysis, and insulin-like growth factor signaling, whereas these signatures are markedly reduced in long-term responders[49].

CSF proteomic profiling enables identification of prognostic biomarkers by defining protein signatures that distinguish aggressive from indolent disease. Enrichment of innate immune, proteolytic, and insulin-like growth factor-related signaling pathways has been correlated with poor outcomes, while long-term survivors display inverse expression patterns of these markers, suggesting potential utility for risk stratification[49]. Functionally, CSF from patients with LMD has been shown to activate PI3K/AKT, integrin, B-cell, tumor necrosis factor receptor 2, TGFβ, and oxidative stress pathways in melanoma cells, linking these signatures to tumor survival and progression[49].

Among these pathways, TGFβ appears to play a central role in therapeutic resistance. Elevated CSF TGFβ levels in patients with progressive LMD have been confirmed by ELISA and are associated with resistance to BRAF inhibitors[49]. In functional assays, CSF from poor responders induced tolerance to BRAF inhibition in melanoma cell cultures, demonstrating that the leptomeningeal microenvironment can directly modulate drug sensitivity. Proteomic analyses further implicate coordinated activation of PI3K/AKT, integrin, TGFβ, and oxidative stress pathways in this resistance phenotype, contributing to adaptive signaling and treatment failure[49].

Beyond resistance biology, proteomic changes within the CSF may provide a dynamic biomarker of disease activity. Alterations in protein expression including rising TGFβ levels correlate with clinical progression and therapeutic response, offering a means for real-time disease monitoring[49,81]. High-throughput mass spectrometry now allows serial CSF sampling to track proteomic evolution over time, and candidate biomarkers have been identified that reflect energy metabolism, vascular remodeling, and inflammatory activity[72,81].

The technical feasibility of CSF proteomics and its capacity to detect disease-specific patterns have generated interest in developing a comprehensive CSF proteome database for diagnostic and prognostic use. Such a resource could enable stratification of patients based on their risk for LMD development or progression, facilitate personalized monitoring of treatment response, and identify therapeutic targets. Given the limitations of cytology and imaging, the use of sensitive, specific, and noninvasive proteomic tools represents a promising new frontier for the diagnosis and management of melanoma LMD[82].

Collectively, these findings establish CSF proteomic profiling as a valuable tool for characterizing the biologic drivers of melanoma LMD, identifying mechanisms of therapeutic resistance, and informing prognostic stratification. By capturing functional pathway activity that is not evident from cytology or imaging alone, CSF proteomics may complement molecular and immune profiling for longitudinal disease monitoring and treatment assessment.

RNA and exosome-based approaches: RNA and exosome-based liquid biopsy approaches represent an expanding frontier in the diagnosis and monitoring of melanoma LMD. These methods analyze extracellular vesicle-associated RNAs including primarily microRNAs (miRNAs), small non-coding RNAs, and exosomal mRNAs isolated from CSF. By leveraging the stability and cell-of-origin specificity of exosomal cargo, these assays provide molecular insight into disease presence, progression, and therapeutic response.

miRNA profiling of CSF has identified unique expression signatures that distinguish leptomeningeal metastases from other CNS tumor states. For example, miR-21 is highly enriched in extracellular vesicles from patients with LMD and is functionally implicated in tumorigenesis, supporting its potential as both a diagnostic and prognostic biomarker[83]. Additional miRNAs, such as miR-335-5p and miR-34b-3p, have been validated as LMD-specific markers, with clustering analyses confirming that CSF-derived miRNA profiles differ significantly from those seen in parenchymal brain metastases or primary brain tumors[84]. In lung adenocarcinoma-associated leptomeningeal metastases, elevated CSF miR-7975, miR-7977, and miR-7641 levels correlate with disease burden and relapse, suggesting potential value for longitudinal monitoring[85].

Exosomal mRNA profiling further broadens the diagnostic and mechanistic utility of CSF analysis. Transcriptomic studies of exosomal messenger RNA in meningeal carcinomatosis, including melanoma LMD, have identified differentially expressed genes involved in oncogenic and microenvironmental signaling, including the PI3K/AKT, integrin, B-cell activation, tumor necrosis factor receptor 2, TGFβ, and oxidative stress pathways[49,86]. These transcripts provide insight into disease biology and may serve as candidate biomarkers of treatment resistance and progression. Functionally, such exosomal mRNA signatures reinforce the overlap between the CSF transcriptome and proteome, suggesting that these vesicles play an active role in mediating tumor-microenvironment communication[86].

Exosomal surface markers have also emerged as valuable diagnostic targets. Tetraspanins such as CD9, CD63, and CD81 are consistently present on CSF-derived exosomes and provide a practical means of identifying and quantifying tumor-associated vesicles[87]. These proteins play a central role in exosome formation and cargo organization, and their expression patterns in CSF differ from those seen in peripheral blood, consistent with a CNS-specific cellular origin[88]. Notably, exosomes isolated from CSF closely resemble brain-derived extracellular vesicles, supporting their relevance to LMD biology. Among these markers, CD81 typically exhibits the strongest signal in LMD, enabling sensitive detection and longitudinal monitoring[87]. Combining markers, such as CD81 with CD63 or CD9, further improves specificity by helping distinguish tumor-derived exosomes from background vesicle populations[89].

Collectively, RNA and exosome-based analyses extend the capabilities of CSF liquid biopsy beyond cytology and cfDNA. They capture the dynamic molecular interactions occurring in the leptomeningeal compartment, provide insight into disease evolution and drug resistance, and offer new opportunities for minimally invasive, high-resolution monitoring of melanoma LMD.

Risk of false positives: Although CSF liquid biopsy technologies offer significant advances in sensitivity and molecular resolution compared with traditional diagnostic methods, interpretation of results requires caution in specific clinical contexts. False-positive findings have been documented in patients with parenchymal brain metastases directly adjacent to CSF spaces. In one study, all three patients with parenchymal tumors adjacent to CSF demonstrated detectable tumor-derived mutations in CSF cell-free DNA despite lacking radiographic, cytologic, or clinical evidence of leptomeningeal involvement[17]. When these cases were excluded, the specificity of CSF-cfDNA analysis substantially improved, and five additional patients with parenchymal tumors not contacting CSF showed uniformly negative results, indicating that proximity to CSF drives this phenomenon. A larger melanoma LMD series reported isolated cfDNA positivity in only one of 85 patients, approximately one percent, and this individual also had an active metastasis bordering the CSF[70]. Notably, six other patients with active brain metastases but no CSF-adjacent disease had negative cfDNA assays. Taken together, these data indicate that CSF liquid biopsy results should be interpreted cautiously in patients with parenchymal metastases abutting CSF spaces and always integrated with imaging and cytologic findings.

MANAGEMENT OF MELANOMA LMD

Historical approaches

Before the development of modern targeted and immune therapies, melanoma LMD was managed primarily with IT chemotherapy, radiation therapy, and symptom-directed neurosurgical interventions, with median survival typically measured in weeks to a few months[1,36]. IT chemotherapy constituted the central treatment modality, most commonly using methotrexate, cytarabine (including liposomal formulations), and thiotepa[90,91]. Ventricular administration via implanted Ommaya reservoir was preferred over lumbar puncture because it provided more reliable CSF distribution and greater patient comfort, with pharmacokinetic studies showing suboptimal delivery in 10%-15% of lumbar injections despite CSF return[91]. Methotrexate (12.5-15 mg weekly) and liposomal cytarabine were commonly used during induction, with some protocols transitioning to maintenance dosing[25,90-92]. However, therapeutic benefit was modest. Retrospective series reported median overall survival of 9.72 weeks in treated patients vs 4.72 weeks without treatment (P = 0.009), and systematic reviews showed partial or complete response rates in only 45% of patients[36,93]. Notably, with significant treatment-related neurotoxicity including meningitis, leukoencephalopathy, seizures, and permanent myelosuppression occurred.

Radiation therapy was similarly used with palliative intent rather than durable disease control. Involved-field radiotherapy targeted symptomatic or radiographically bulky regions of disease, while whole-brain radiation therapy (WBRT) was recommended for palliation when focal techniques were not feasible[22,92]. Contemporary dosing strategies and hippocampal-sparing techniques are discussed in he section “Radiation innovations” below. Retrospective analyses suggested modest survival benefit. For example, patients receiving WBRT had median survival of 15.72 weeks compared with 7.88 weeks in those who did not (P = 0.0007)[36]. Combined chemoradiotherapy using weekly IT methotrexate with 40 Gy of involved-field radiotherapy achieved an 86% response rate and a median survival of 6.5 months in mixed-tumor LMD cohorts, though treatment was associated with significant grade III-V toxicity[92]. Ultimately, historical multimodal approaches that combined ventricular reservoir placement, radiotherapy to improve CSF circulation, and serial IT chemotherapy were used mainly to relieve symptoms and slow neurological deterioration. Despite aggressive intervention, survival generally remained less than three months for patients in historical series, and no randomized trial ever demonstrated a definitive survival advantage for IT chemotherapy over optimal systemic therapy or best supportive care[36,91].

Systemic therapies

Systemic immune checkpoint inhibition forms the backbone of contemporary management for melanoma LMD, although clinical outcomes are still inferior to those seen in parenchymal MBM. Combination immune checkpoint blockade with nivolumab and ipilimumab leads to the highest intracranial response rates in metastatic melanoma, reaching 46%-55% in patients with asymptomatic brain metastases[94,95]. However, most patients with LMD present with symptomatic neurological deficits and require corticosteroids, which diminish ICIs effectiveness[95]. Data specific to LMD remain limited, though a phase II trial of intravenous (IV) ipilimumab plus nivolumab in leptomeningeal carcinomatosis from multiple solid tumors reported a 3-month overall survival rate of 44%, meeting its predefined efficacy endpoint[96]. In contrast, patients with leptomeningeal metastases in the ABC trial progressed quickly on nivolumab, with a 3-year intracranial progression-free survival of only 6%, underscoring the highly aggressive nature of LMD[94].

Single-agent PD-1 inhibitors such as pembrolizumab or nivolumab demonstrate only modest intracranial activity (about 20% response rates) and are generally not preferred as initial monotherapy for CNS disease[97,98]. Even so, they remain useful in selected patients, particularly those who cannot tolerate combination ICI therapy. In individuals with BRAF V600-mutant melanoma, targeted therapy with BRAF/MEK inhibitor combinations such as dabrafenib-trametinib, vemurafenib-cobimetinib, or encorafenib-binimetinib can produce substantial intracranial activity, with response rates in brain metastases reported as high as 58%[95,97]. However, durability of response is typically shorter in the CNS compared to extracranial sites[97]. National Comprehensive Cancer Network (NCCN) guidelines endorse BRAF/MEK inhibition particularly in symptomatic patients or those with substantial extracranial disease who have not previously received targeted therapy[97,99]. Importantly, concurrent use of BRAF/MEK inhibitors was permitted in the IT/IV nivolumab trials, suggesting compatibility with emerging leptomeningeal-directed immunotherapy strategies[12,100].

Overall, systemic therapies provide measurable but limited benefit in melanoma LMD. A nationwide Danish cohort similarly reported a median overall survival of 8.4 months and found no survival difference between patients treated with ICIs and those receiving BRAF/MEK inhibitors[11]. While combination ICI and targeted therapy have transformed outcomes for parenchymal CNS disease, their efficacy in LMD is lower, underscoring the need for therapeutic approaches that overcome the unique immune barriers of the leptomeningeal compartment. Clinical decision-making in melanoma LMD is guided by the European Society for Medical Oncology-European Association of Neuro-Oncology leptomeningeal metastasis guidelines which emphasize patient stratification based on performance status, neurologic involvement, CSF dynamics, and systemic disease burden to inform the use of systemic therapy, IT treatment, radiotherapy, or supportive care[101].

IT immunotherapy (emerging frontier)

IT immunotherapy has rapidly evolved into one of the most promising strategies for melanoma LMD as it addresses the restricted penetration of systemic ICIs across the blood-CSF barrier. The most robust evidence comes from a phase I/Ib trial of concurrent IT and IV nivolumab (NCT03025256), which established IT nivolumab 50 mg every two weeks (administered via Ommaya reservoir) combined with IV nivolumab 240 mg as the recommended regimen[12,100]. The regimen was well tolerated, with no dose-limiting toxicities identified across the 5 mg to 50 mg IT dose levels. Grade 3 adverse events occurred in 18% of patients, and no grade 4 or 5 toxicities were reported[100]. Rash (58%), nausea (44%), vomiting (34%), and dizziness (22%) were the most common treatment-related events, while IT effects were usually mild with only one patient developing grade 3 vasogenic edema[100].

Clinical efficacy signals from the 50 patients dose escalation and expansion cohorts are highly encouraging compared to historical outcomes. Median overall survival reached 7.0-7.5 months, with survival rates of 66.6% at 3 months, 53.1% at 6 months, and 34.8% at 12 months[12,100]. Importantly, 84% of patients had received prior anti-PD-1 therapy, yet IT/IV nivolumab still demonstrated activity[100]. These results are a significant improvement over the historical median survival of less than 3 months in melanoma LMD treated with conventional modalities. Treatment was feasible in patients receiving low-dose corticosteroids (≤ 4 mg dexamethasone equivalent daily) or concurrent BRAF/MEK inhibitors[100,102]. Based on this, the NCCN now lists IT/IV nivolumab as a “useful in certain circumstances” option (category 2B), marking the first IT immunotherapy to receive guideline acknowledgment for melanoma LMD[99].

The biological rationale for IT checkpoint blockade is supported by mechanistic studies demonstrating that direct administration into CSF achieves higher leptomeningeal drug concentrations than systemic dosing alone, overcoming a major pharmacologic barrier to treating LMD[12,103]. Preclinical models show that dual-compartment PD-1 blockade (IT plus systemic) significantly improves survival compared to monotherapy through enhanced local immune activation and systemic disease control[104]. Building on these findings, an ongoing phase Ib expansion is testing IT nivolumab plus IT relatlimab, a lymphocyte-activation gene 3 (LAG-3) inhibitor. This strategy was prompted by detection of LAG-3 on CSF-resident CD8 T cells in patients with LMD and by animal studies showing that adding LAG-3 inhibition to PD-1 blockade enhances antitumor activity[105]. Additional IT immunotherapy modalities including IT pembrolizumab, IL-2, tumor-infiltrating lymphocytes, CAR-T cells, and catheter-based delivery systems are under investigation. Early retrospective experience with continuous IT infusion (10-13.5 mg/day) suggests that this approach may be less effective than intermittent bolus dosing[106].

Collectively, IT immunotherapy represents a major advancement for melanoma LMD, offering extended survival, a favorable safety profile, and the potential to integrate with other targeted and immune therapies. Ongoing clinical trials will determine optimal combinations and sequencing strategies to further improve outcomes.

Radiation innovations

Radiation therapy for melanoma LMD has historically been limited to palliative approaches such as WBRT or involved-field radiotherapy, but recent advances have introduced strategies capable of achieving broader neuroaxis control with reduced toxicity. The most important development is pCSI. In a randomized phase II study, patients who received pCSI had longer CNS progression-free survival (7.5 months vs 2.3 months) and overall survival (9.9 months vs 6.0 months; P = 0.029) compared with those treated with involved-field radiotherapy, prompting early closure of the trial because of clear efficacy[14,107]. Because protons deposit most of their energy at the Bragg peak with minimal exit dose, proton CSI limits exposure to bone marrow and abdominal organs and reduces radiation delivered beyond the neuroaxis[4,14]. The NCCN now recommends proton-based techniques for craniospinal irradiation when available, emphasizing bone marrow-sparing approaches for metastatic solid tumors involving the CNS[99]. WBRT remains appropriate primarily for palliative care in patients who are not candidates for SRS and who exhibit radiographic or clinical evidence of LMD. Typical doses are 30 Gy in 10 fractions or 20 Gy in 5 fractions. For patients with an anticipated survival of four months or more and no hippocampal involvement, hippocampal-avoidance WBRT with memantine is recommended to reduce neurocognitive toxicity[97,99]. Although SRS is the preferred approach for focal parenchymal melanoma metastases, its application in LMD is limited by the diffuse pattern of spread. It remains an option for isolated bulky or symptomatic nodular lesions[14,97].

Beyond external-beam radiation techniques, radionuclide therapeutics have emerged as a potential, though still experimental strategy. Rhenium-186 nanoliposomes (186RNL) represent a developing radiotheranostic platform capable of delivering high-dose beta radiation via convection-enhanced delivery. Basically, this approach uses radioactive nanoparticles infused directly into the tumor to deliver highly concentrated radiation, while minimizing exposure to surrounding tissue. A phase I study in recurrent glioma found that doses up to 22.3 mCi were well tolerated, and median overall survival was 11 months; patients who received absorbed doses above 100 Gy lived longer on average, about 17 months vs 6 months in those receiving lower doses[108]. 186RNL accumulates in tumors through the enhanced permeability and retention effect, and its rhenium-186 component acts as the beta-emitting source[109,110]. Although these early data show that targeted radionuclide therapy is technically feasible, there is no clinical evidence supporting the use of 186RNL in melanoma LMD, and major obstacles remain, including achieving even distribution in the CSF and delivering adequate dose along the neuroaxis. As systemic and IT therapies advance, pairing them with newer radiation techniques such as pCSI or investigational radionuclide approaches may open the door to future treatment strategies for LMD.

Neurosurgical considerations

Neurosurgical interventions in melanoma LMD primarily support diagnosis, enable IT drug delivery, and relieve symptoms related to hydrocephalus and intracranial hypertension. Surgical biopsy is rarely required because the diagnosis is typically established by CSF cytology and MRI, with neurosurgical procedures primarily reserved for CSF diversion or symptom palliation[111]. When lumbar puncture is contraindicated or repeatedly nondiagnostic despite strong clinical suspicion, placing a ventricular reservoir can help obtain CSF more reliably[4,22,112]. CSF diversion is the most common neurosurgical intervention, with ventriculoperitoneal shunting (VPS) and Rickham reservoir (RR) placement providing symptomatic improvement in 79%-83% of patients[113-115]. VPS achieves symptom relief in approximately 79% of cases with median post-shunt survival of 2.4-3.8 months[113,116]. RR systems have lower revision rates (8% vs 24% with VPS) and similar survival outcomes, making initial RR placement with later VPS conversion a reasonable strategy for select patients[115]. Complications include shunt malfunction or revision, infection, and subdural collections, though abdominal tumor seeding via VPS has not been a significant concern[112,113]. Patients with cerebellar enhancement or swelling experience shorter survival and may derive less benefit, whereas those with supratentorial edema, including disproportionately enlarged subarachnoid space hydrocephalus, may respond more favorably to diversion[117].

IT therapy delivery is another key neurosurgical function. Ommaya reservoirs, small devices implanted under the scalp and connected to a ventricular catheter, enable reliable intraventricular access for repeated IT chemotherapy without the challenges of serial lumbar punctures[93,112]. For patients requiring both CSF diversion and IT therapy, reservoir-on/off valve VPS systems permit IT drug administration while maintaining CSF drainage. In one series using this system, 83.3% of patients experienced symptomatic improvement and 61.1% achieved cytologic responses[118]. These approaches have become increasingly relevant with the expansion of IT immunotherapy. Symptomatic intracranial hypertension occurs in roughly 58% of patients, and CSF diversion results in meaningful gains: 79%-83% experiencing symptom relief, 79% are discharged home or to rehabilitation, and 56% receive additional systemic or radiation therapy[113,114,119].

In addition to diagnostic and therapeutic roles, neurosurgical interventions and multidisciplinary palliative care are essential for managing hydrocephalus, neurologic deficits, and overall symptom burden in advanced melanoma LMD. Comprehensive palliative care, in particular, is essential for addressing neurocognitive decline, encephalopathy, hydrocephalus-related symptoms, and systemic complications, and should be integrated early alongside disease-directed therapy. Neurosurgical decisions should consider the expected quality-of-life benefit, the patient’s goals, and whether the intervention will allow further oncologic treatment, as outcomes are generally better in patients with higher performance status and lower metastatic burden[113,120].

CURRENT CHALLENGES

Despite recent advances, melanoma LMD remains associated with substantial unmet clinical needs. One of the most significant barriers to progress is the historical exclusion of patients with LMD from clinical trials. Many pivotal studies of ICIs, targeted therapies, and CNS-directed treatments have excluded patients with leptomeningeal involvement because of poor prognosis, neurologic symptoms, or corticosteroid use. As a result, evidence guiding treatment decisions in LMD is largely derived from small cohorts, retrospective analyses, or extrapolation from parenchymal brain metastasis studies, limiting the strength and generalizability of current recommendations.

Even when responses are achieved, durability remains limited. The leptomeningeal compartment presents a uniquely hostile therapeutic environment characterized by restricted drug penetration, altered CSF flow dynamics, and a profoundly immunosuppressive immune landscape. While single-cell RNA sequencing has clarified the immunosuppressive composition of the CSF, how melanoma cells interact with and shape these immune populations remains largely undefined. These factors contribute to early progression and acquired resistance across systemic, intrathecal, and radiation-based approaches. While IT immunotherapy and pCSI have extended survival in selected patients, most responses remain transient.

Finally, the field lacks standardized diagnostic criteria, response metrics, and clinical endpoints tailored to LMD. Diagnostic and response assessment methods vary widely across studies, ranging from cytology and imaging to liquid biopsy and clinical evaluation, making results difficult to compare. Standard oncologic endpoints, including radiographic response and progression-free survival, often fail to capture meaningful neurologic outcomes in diffuse LMD. The absence of uniform diagnostic and outcome frameworks continues to hinder trial design, regulatory evaluation, and the development of evidence-based treatment algorithms.

FUTURE DIRECTIONS AND RESEARCH PRIORITIES

Future progress in melanoma LMD will depend on sustained expansion of clinical research specifically designed for this population. Although patients with LMD have historically been excluded from most melanoma trials, recent early-phase studies now directly enroll individuals with leptomeningeal involvement, including trials of IT and combined intrathecal-systemic immunotherapy. These efforts represent an important shift, but they remain limited in number, scope, and phase. Broader inclusion of LMD patients in prospective studies, along with trial designs that accommodate neurologic symptoms and corticosteroid use, will be essential to generate more generalizable data.

Therapeutically, rational combination strategies are one of the most promising avenues to improve outcomes. These strategies include combining systemic ICIs with IT therapy, pairing immunotherapy with targeted agents in BRAF-mutant disease, and integrating systemic or IT treatments with CNS-directed radiation. Beyond PD-1-based regimens, future immunotherapeutic strategies may also involve alternative checkpoint targets such as LAG-3 or T-cell immunoreceptor with Ig and ITIM domains, as well as interventions aimed at modulating myeloid-driven immunosuppression within the CSF. Early clinical experience suggests that treating both the systemic and leptomeningeal compartments may help address barriers to drug delivery and immune activity within the CSF, although the best sequencing, dosing, and patient selection strategies remain unclear. Continued investigation into resistance mechanisms within the leptomeningeal microenvironment will be critical to guide these strategies.

Advances in diagnostics and preclinical modeling are also likely to shape the next phase of LMD research. Further refinement of CSF liquid biopsy platforms, including cell-free DNA, CTC, and proteomic analyses, may enable real-time monitoring of disease burden, therapeutic response, and emerging resistance which would then support more adaptive treatment strategies. Progress will also depend on better preclinical models. Patient-derived CSF CTC xenografts and murine platforms that allow intraventricular drug delivery, including Ommaya-like systems, offer practical tools to study drug penetration, resistance, and treatment sequencing. These models could help guide development of therapies designed to function within the CSF, including engineered antibodies, nanoparticle-based delivery approaches, and cellular therapies.

Finally, progress in melanoma LMD will require closer integration of diagnostic, biologic, and clinical endpoints. Traditional radiographic response criteria are often insufficient in a disease characterized by diffuse involvement and neurologic decline. Future trials will likely need to incorporate composite endpoints that integrate clinical status, CSF-based biomarkers, and longitudinal molecular profiling. Such approaches may better capture meaningful benefit, enable earlier assessment of therapeutic response, and support adaptive trial designs tailored to the unique biology of LMD.

CONCLUSION

Melanoma LMD has long been one of the most serious complications of advanced melanoma, often leading to rapid neurological decline and few treatment options. In the past ten years, there have been important advances. New CSF diagnostics have improved detection and allowed for more detailed molecular analysis. Modern systemic therapies, IT immunotherapy, and CNS-directed radiation strategies have modestly extended survival beyond historical benchmarks in selected patients. Nevertheless, melanoma LMD continues to be associated with uniformly high mortality and remains largely refractory to available treatments.

Continued progress will rely on focused, disease-specific efforts. Including more patients in clinical trials, using standard diagnostic and response criteria, and developing treatments for the leptomeningeal area are key next steps. As we learn more about the biology and run more clinical trials, melanoma LMD is becoming a target for new therapies rather than a diagnosis with few options. Working together across specialties will be important for lasting improvements in patient care.