Outcomes and prognostic factors in melanoma patients with leptomeningeal disease: a retrospective cohort study

Abstract

Leptomeningeal disease (LMD) is a devastating complication of metastatic melanoma, affecting up to 15% of patients with central nervous system involvement. Despite advances in systemic therapy, LMD carries poor prognosis. Patients with LMD are often excluded from clinical trials. Contemporary real-world data on prognostic factors and treatment outcomes in melanoma LMD are limited. We conducted a retrospective cohort study of patients diagnosed with leptomeningeal metastasis from melanoma between 1 January 2018 and 31 December 2025, across Mayo Clinic sites. LMD diagnosis was made by magnetic resonance imaging showing leptomeningeal enhancement and confirmation from cerebrospinal fluid cytology and clinical presentation. Demographic, clinical, treatment, and outcome data were collected. Overall survival (OS) from LMD diagnosis was the primary endpoint. Forty-four patients were identified (median age: 60.0 years, range: 30.5–82.7). Before diagnosis of LMD, systemic therapy for metastatic disease was given to 91% of patients. BRAF mutations were present in 65.9%. After LMD diagnosis, 88.6% were treated for LMD. Median OS was 2.96 months (95% confidence interval: 1.39–4.53). Poor Eastern Cooperative Oncology Group performance status was independently associated with worse OS (hazard ratio: 2.10, P = 0.036). Any treatment for LMD (hazard ratio: 0.16, P = 0.005) was associated with improved OS. Melanoma-related LMD remains rapidly fatal. Studies of novel treatments are needed to improve outcomes for patients with this diagnosis.

Introduction

Leptomeningeal disease (LMD), also referred to as carcinomatosis meningitis or neoplastic meningitis, is a severe and often terminal complication of advanced malignancies characterized by the dissemination of cancer cells to the leptomeninges (pia and arachnoid mater) and the cerebrospinal fluid (CSF) within the subarachnoid space [1,2]. Malignant melanoma is one of the most common solid tumors to cause LMD, with incidence estimates ranging from 10 to 15% [1,3]. Autopsy studies suggest central nervous system (CNS) involvement in up to 75% of patients with metastatic melanoma, with LMD detected in as many as 20% [2,4]. The incidence of melanoma-related LMD is increasing, likely due to improved imaging techniques, longer survival from systemic therapies, and heightened surveillance protocols [1–3].

Despite advances in the treatment of metastatic melanoma, outcomes for patients with LMD remain poor [1,2]. The pathophysiology of LMD is not fully understood, but hematogenous spread, direct extension from parenchymal brain metastases, or perineural invasion have all been proposed as potential mechanisms of leptomeningeal metastasis [3]. Although CSF is typically nutrient-poor and inhospitable to tumor cells, it can become permissive to melanoma growth, partly mediated by complement component 3 (C3), which disrupts the blood-CSF barrier and enhances melanoma cell invasion [1,2]. Moreover, recent findings into the immune landscape of LMD from single-cell RNA sequencing studies have identified profound T cell exhaustion, enrichment of myeloid-derived suppressor cells, and a scarcity of functional CD8+ T cells, contributing to a highly immunosuppressive microenvironment [2–4].

Diagnosis of LMD is challenging; CSF cytology remains the gold standard but carries a false-negative rate of up to 50% [5–7], while MRI findings can be nonspecific and prone to false positives [2,3]. Emerging methods, such as CSF circulating tumor cells and cell-free tumor DNA, may offer improved sensitivity and prognostic value and could help track treatment response in the future but are not currently available for widespread clinical use [5,7].

Treatment options for melanoma LMD are limited and often palliative. Systemic therapies, such as immune checkpoint inhibitors (ICIs) and BRAF/MEK-targeted agents, constitute the foundation of care for metastatic melanoma, but their efficacy within the leptomeningeal space is often hindered by poor CSF penetration [1–3]. Radiation therapy, including whole-brain radiation, may help alleviate neurologic symptoms but has not been consistently shown to improve overall survival (OS) [8–10]. Intrathecal therapies delivered via Ommaya reservoirs have shown modest activity and are often accompanied by significant toxicity [2,3,11]. Consequently, median OS for melanoma LMD remains dismal, ranging from 1.6 to 4.9 months across various retrospective cohorts [3,12,13], with one study reporting a median OS of just 1.8 months for those treated with intrathecal interleukin-2 [14] and another achieving 4.9 months in a post-2015 cohort [15].

There remains a substantial gap in prospective clinical trials and standard-of-care protocols for melanoma LMD. Most clinical studies have excluded these patients due to poor performance status and neurologic compromise [1–3]. Emerging data on concurrent intrathecal and intravenous immunotherapy, as well as improved molecular diagnostics, have underscored the feasibility of more aggressive or tailored approaches in select patients [4,16,17]. Still, more robust data are needed to guide treatment in real-world practice.

This study aimed to evaluate clinical outcomes, treatment patterns, and prognostic factors in a contemporary cohort of melanoma patients with LMD treated across Mayo Clinic sites. By leveraging retrospective data, we sought to identify survival trends, assess the impact of systemic and radiation therapies, and explore patient characteristics associated with improved outcomes, contributing to the limited but growing body of literature guiding the management of this vulnerable population.

Methods

Study design and purpose

We conducted a retrospective cohort study to describe the clinical characteristics, treatment patterns, and outcomes of patients with LMD secondary to melanoma who were diagnosed or treated at Mayo Clinic sites in Minnesota, Arizona, Florida, and community sites of Mayo Clinic Health System. The study was approved by the institutional review board at Mayo Clinic.

Patient identification and eligibility

Patients were identified through the Mayo Clinic Data Explorer tool, a centralized clinical data query platform that enables structured searches across institutional databases. The study included patients diagnosed with LMD secondary to melanoma between 1 January 2018 and 31 December 2025.

Initial identification was performed using relevant International Classification of Diseases diagnostic codes, including C79.32 (secondary malignant neoplasm of meninges), C79.31 (secondary malignant neoplasm of brain), and melanoma-related codes (e.g. C43.x). Records were then reviewed by members of the research team to confirm that each case met predefined inclusion criteria. In accordance with institutional policy and Minnesota state law, Minnesota Research Authorization was confirmed for all patients from Minnesota prior to inclusion.

To be included, patients had to have: a confirmed diagnosis of melanoma as the primary malignancy, and either MRI or cytological evidence of leptomeningeal disease. MRI findings were either reported as consistent with leptomeningeal disease by the interpreting radiologist or deemed highly suspicious and confirmed by the treating physician. Given the retrospective nature of the study, neither imaging nor CSF samples were re-reviewed by study investigators.

Although CSF cytology and clinical findings (e.g. headache, nausea, vomiting, seizures, cognitive/behavioral changes, gait instability, cranial neuropathies, or focal motor/sensory deficits) were reviewed when available, all included patients had MRI findings consistent with leptomeningeal disease. In clinical practice, MRI findings typically prompt further evaluation, with CSF obtained in selected cases to support the diagnosis. LMD was not uniformly symptom-driven and was incidentally identified on routine surveillance imaging in asymptomatic patients.

Patients were excluded if LMD was diagnosed solely at autopsy, if the primary malignancy was not melanoma, or if follow-up data were unavailable.

Data collection

Demographic, clinical, pathologic, laboratory, and treatment data were extracted from the electronic medical record and entered into a secure REDCap database. Collected variables included patient demographics, melanoma characteristics, prior melanoma-directed therapies, details of LMD diagnosis, and performance status. Laboratory values, such as serum lactate dehydrogenase (LDH), were recorded when available.

Post-LMD treatment data included systemic therapy (including ICIs, BRAF/MEK-targeted therapies, and chemotherapy), radiation therapy (whole-brain, spinal, or stereotactic radiosurgery), neurosurgical interventions, and concurrent corticosteroid use during systemic therapy.

Outcomes

The primary outcome was OS, defined as the time from LMD diagnosis to death from any cause or last follow-up. Secondary outcomes included progression-free survival (PFS) among patients who received systemic therapy for LMD, defined as time from initiation of systemic treatment to clinical or radiographic progression or death. Time intervals from initial melanoma diagnosis to CNS progression and from CNS progression to LMD diagnosis were also recorded.

Statistical analysis

Descriptive statistics were used to summarize patient characteristics, treatment patterns, and outcomes. Continuous variables were reported as medians with interquartile ranges (IQR), and categorical variables as frequencies with percentages. Patients with missing data for a given variable were excluded from that specific analysis; no imputation was performed. Survival outcomes were estimated using the Kaplan–Meier method. Hazard ratios and 95% confidence intervals (CIs) were calculated using univariate and multivariate Cox proportional hazards models to identify prognostic factors associated with OS. Variables significant on univariate analysis (P < 0.05) were included in the multivariate model. A two-sided P value less than 0.05 was considered statistically significant. Statistical analyses were conducted using IBM SPSS Statistics, version 28 (IBM Corp., Armonk, New York, USA).

Results

Patient characteristics

A total of 44 patients with melanoma LMD were included. The median age at initial melanoma diagnosis was 54.6 years (range, 14.6–79.6; IQR, 24.9) and 60.0 years at LMD diagnosis (range, 30.5–82.7; IQR, 27.2). The cohort comprised 26 male (59.1%) and 18 female (40.9%) patients. BRAF mutations were present in 29 patients (65.9%), 14 (31.8%) were BRAF wildtype, and 1 (2.3%) had unknown mutation status. Primary melanoma subtypes included cutaneous (84.1%), mucosal (4.5%), uveal (2.3%), and unknown primary (9.1%).

CSF cytology was positive in 4 patients (9.1%), negative in 3 (6.8%), atypical in 9 (20.5%), and not performed in 28 patients (63.6%). All patients had MRI-confirmed LMD, and a suggestive clinical picture was documented in 34 patients (77.3%). All patients had stage IV melanoma prior to LMD diagnosis, with 35 patients (79.5%) having CNS metastases (AJCC M1d) and 6 patients (13.6%) having extracranial visceral metastases (M1c), while only 3 patients (6.8%) had limited metastatic disease (M1a–M1b). Elevated LDH was observed in 12 patients (27.3%), normal in 30 (68.2%), and not available in 2 (4.5%). Prior systemic therapy for metastatic disease had been given to 40 patients (91.0%), including ICIs (35 patients, 79.5%), BRAF/MEK-targeted therapy (24, 54.5%), chemotherapy (7, 15.9%), and greater than or equal to 2 drug classes (7, 15.9%). Prior CNS-directed radiation was reported in 25 patients (56.8%), and prior surgical resection in 14 (31.8%). Baseline demographics and clinical characteristics are summarized in Table 1.

Table 1.

Summary of baseline characteristics and treatment modalities in melanoma patients with leptomeningeal disease (N = 44)

Characteristic	N (%) or value	Median (range/IQR or 95% CI)
Demographics
Age at initial melanoma diagnosis	–	54.6 (14.6–79.6/24.9) years
Age at LMD diagnosis	–	60.0 (30.5–82.7/27.2) years
Sex
Male	26 (59.1%)
Female	18 (40.9%)
Tumor features
Mutation status
BRAF-mutant	29 (65.9%)
Wildtype	14 (31.8%)
Unknown	1 (2.3%)
Primary melanoma subtype
Cutaneous	37 (84.1%)
Mucosal	2 (4.5%)
Uveal	1 (2.3%)
Unknown primary	4 (9.1%)
Diagnosis/labs
CSF cytology
Positive	4 (9.1%)
Negative	3 (6.8%)
Atypical cells	9 (20.5%)
Not done	28 (63.6%)
ECOG performance status
Good (0–1)	22 (50%)
Poor (2–4)	17 (38.6%)
Unknown	5 (11.4%)
Clinical impression of LMD
Yes	34 (77.3%)
No	10 (22.7%)
LDH > ULN at LMD diagnosis
Yes	12 (27.3%)
No	30 (68.2%)
Not available	2 (4.5%)
Prior treatment history
Previous systemic therapy (any)	40 (91.0%)
Immunotherapy	35 (79.5%)
Targeted therapy	24 (54.5%)
Chemotherapy	7 (15.9%)
≥2 classes (sequential)	7 (15.9%)
Previous CNS radiation
Yes	25 (56.8%)
No	19 (43.2%)
Previous surgical resection (CNS)
Yes	14 (31.8%)
No	30 (68.2%)
Clinical outcomes
Time from initial melanoma to CNS progression	–	36.1 months (0.0–265.4/ 87.3)
Time from CNS progression to LMD diagnosis	–	5.8 months (0.0–103.6/ 19.5)
PFS	–	45.0 days (22.8–67.2, 95% CI)
Median OS	–	2.96 months (1.39–4.53, 95% CI)
Treatment modalities
Any treatment for LMD	39 (88.6%)
Radiation (any)	31 (70.5%)b
Whole-brain RT	19 (43.2%)
Spinal RT	6 (13.6%)
SRS	4 (9.1%)
Craniospinal radiation	3 (6.8%)
Proton craniospinal radiation	1 (2.3%)
Systemic therapy (any)	32 (72.7%)
ICI	18 (40.9%)
Nivolumab and ipilimumab	10 (22.7%)
Pembrolizumab	5 (11.4%)
Nivolumab only	1 (2.3%)
Pembrolizumab and ipilimumab	1 (2.3%)
BRAF/MEK-targeted	19 (43.2%)
Dabrafenib and trametinib	11 (25%)
Binimetinib and encorafenib	7 (15.9%)
Vemurafenib and cobimetinib	2 (4.5%)
On steroids during ICI or BRAF/MEKa	20 (62.5%)
Chemotherapy	3 (6.8%)
Both ICI and targeted	5 (11.4%)
No treatment	5 (11.4%)
Treatment group
Only radiation	7 (15.9%)
Only systemic	8 (18.2%)
Both radiation and systemic	24 (54.5%)
ICI and radiation	13 (29.5%)
BRAF/MEK and radiation	11 (25%)

Percentages are calculated out of the total cohort (N = 44), unless otherwise specified.

BRAF, B-Raf proto-oncogene; CNS, central nervous system; CSF, cerebrospinal fluid; ECOG, Eastern Cooperative Oncology Group; ICI, immune checkpoint inhibitor; LMD, leptomeningeal disease; MEK, mitogen-activated protein kinase; OS, overall survival; PFS, progression-free survival; RT, radiotherapy; SRS, stereotactic radiosurgery; ULN, upper limit of normal.

^aPercentage among those treated with ICI or BRAF/MEK therapy.

^bSome patients received more than 1 modality of radiation therapy.

Treatment after leptomeningeal disease diagnosis

Following diagnosis of LMD, 39 patients (88.6%) received treatment for LMD. Radiation therapy was administered to 31 patients (70.5%), of which were whole-brain radiation (19, 43.2%), spinal radiation (6, 13.6%), stereotactic radiosurgery (4, 9.1%), craniospinal radiation (3, 6.8%), and proton craniospinal radiation (1, 2.3%). Systemic therapy was given to 32 patients (72.7%). ICIs were used in 18 patients (40.9%), most commonly the combination of nivolumab and ipilimumab (10, 22.7%), followed by pembrolizumab (5, 11.4%), nivolumab monotherapy (1, 2.3%), and one patient who received both pembrolizumab and ipilimumab (1, 2.3%). BRAF/MEK-targeted therapy was administered to 19 patients (43.2%), with dabrafenib and trametinib being the most frequently used regimen (11, 25.0%), followed by encorafenib and binimetinib (7, 15.9%), and vemurafenib and cobimetinib (2, 4.5%). Chemotherapy was used in 3 patients (6.8%). Concurrent ICIs and BRAF/MEK-targeted therapy were given to 5 patients (11.4%). Of the 32 patients treated with ICIs or BRAF/MEK inhibitors, 20 (62.5%) received corticosteroids. 13 patients (29.5%) received concurrent ICI and radiation, and 11 (25%) received simultaneous BRAF/MEK-targeted therapy and radiation.

Overall, 24 patients (54.5%) received both radiation and systemic therapy, 7 (15.9%) received radiation only, 8 (18.2%) received systemic therapy only, and 5 (11.4%) received no treatment for LMD. The median number of LMD-specific treatments was 2.0 (range, 0–5; IQR: 2). Detailed treatment modalities are included in Table 1. Treatment modality overlaps are summarized in Fig. 1, which illustrates the distribution of patients receiving radiation, systemic therapy, or both, along with corresponding survival outcomes.

Fig. 1.

Treatment patterns and outcomes for melanoma leptomeningeal disease. Venn diagram showing the distribution of central nervous system–directed radiation and systemic therapy among patients who received either modality. Left circle: radiation (n = 31); right circle: systemic therapy (n = 32). Segments denote mutually exclusive groups: radiation only (n = 7), systemic therapy only (n = 8), BRAFi/MEKi + radiation (n = 11), and ICI + radiation (n = 13). Patients who received neither modality are not shown. Numbers within the diagram indicate patient counts. Median overall survival (months, 95% CI, log-rank P value): radiation only: 4.11 (0.65–7.57), P = 0.443; systemic therapy only: 3.06 (1.87–4.25), P = 0.559; BRAFi/MEKi + radiation: 3.55 (1.77–5.33), P = 0.661; and ICI + radiation: 1.91 (0.00–3.96), P = 0.291. Created in BioRender. Rezazadeh Roudkoli, A. (2025) https://BioRender.com/9o24ct7. CI, confidence interval; ICI, immune checkpoint inhibitor; LMD, leptomeningeal disease; OS, overall survival.

Clinical outcomes

The median time from initial melanoma diagnosis to CNS progression was 36.1 months (range: 0.0–265.4; IQR: 87.3), and the interval from CNS progression to LMD diagnosis was 5.8 months (range: 0.0–103.6; IQR: 19.5).

Among the 32 patients who received systemic therapy for LMD, median PFS was 1.5 months (95% CI: 0.76–2.24). Median overall survival (OS) for the entire cohort was 2.96 months (95% CI: 1.39–4.53; Fig. 2a). Kaplan–Meier survival curves for the overall cohort and relevant subgroups are shown in Fig. 2.

Fig. 2.

Kaplan–Meier curves for OS in melanoma patients with leptomeningeal disease (LMD): (A) OS for entire cohort; (B) OS by ECOG performance score (HR 2.10, P = 0.036); (C) OS by receipt of any treatment of LMD (HR 0.16, P = 0.005); (D) OS by receipt of ICI and radiotherapy concurrently (HR 0.54, P = 0.102). ECOG, ECOG, Eastern Cooperative Oncology Group; HR, hazard ratio; ICI, immune checkpoint inhibitor; LMD, leptomeningeal disease; OS, overall survival; RT, radiotherapy.

Prognostic factors for overall survival

On univariate analysis, factors significantly associated with shorter OS included lower Eastern Cooperative Oncology Group (ECOG) performance status (hazard ratio 2.30; 95% CI: 1.13–4.63; P = 0.022) and prior CNS surgery (hazard ratio 2.05; 95% CI: 1.04–4.06; P = 0.039). Individuals with parenchymal brain metastases had a trend toward worse survival (hazard ratio 3.74; 95% CI: 0.49–28.69; P = 0.205). LMD treated with ICIs or BRAF/MEK inhibitors requiring steroids had a nonsignificant trend toward reduced OS (hazard ratio 1.95; 95% CI: 0.88–4.31; P = 0.101).

Improved OS was associated with receipt of any treatment for LMD (hazard ratio 0.13; 95% CI: 0.04–0.45; P = 0.001) and receipt of LMD-directed radiation therapy (hazard ratio 0.43; 95% CI: 0.21–0.87; P = 0.019). Concurrent ICI plus radiation (hazard ratio 0.54; 95% CI: 0.28–1.13; P = 0.102) and ICI therapy alone (hazard ratio 0.60; 95% CI: 0.31–1.16; P = 0.125) each had a trend toward improved survival.

On multivariate Cox regression analysis, an independent predictor of worse OS included worse ECOG performance status (hazard ratio 2.10; 95% CI: 1.05–4.46; P = 0.036), while receipt of any therapy for LMD was associated with better survival (hazard ratio 0.16; 95% CI: 0.04–0.57; P = 0.005). Survival by ECOG performance status, receipt of therapy for LMD, and concurrent ICIs plus radiation are shown in Fig. 2b–d.

These findings underscore the prognostic impact of functional status and highlight the survival benefit of LMD-directed interventions in this patient population. Results of the univariate and multivariate Cox regression analyses are presented in Table 2.

Table 2.

Univariate and multivariate analysis of prognostic factors for overall survival in melanoma leptomeningeal disease

Variable	Median OS (month) [95% CI]	KM P value	Univariate HR (95% CI)	Univariate P value	Multivariate HR (95% CI)	Multivariate P value
Sex (male vs female)	2.76 (1.36–4.16)	0.327	1.38 (0.72–2.64)	0.33	–	–
Age at LMD diagnosis (>65 vs ≤65)	2.17 (0.53–3.81)	0.849	1.07 (0.55–2.07)	0.850	–	–
BRAF (mutated vs WT)	2.76 (1.42–4.10)	0.607	1.21 (0.59–2.46)	0.608	–	–
CSF cytology (positive vs atypical cells and negative)	4.34 (0.00–15.21)	0.862	1.13 (0.29–4.36)	0.862	–	–
Extent of LMD (whole CNS vs focal)	2.76 (1.88–3.64)	0.710	1.20 (0.46–3.09)	0.711	–	–
Parenchymal brain metastasis (yes vs no)	2.96 (1.43–4.49)	0.177	3.74 (0.49–28.69)	0.205	–	–
Systemic (non-CNS) disease (yes vs no)	2.76 (1.50–4.02)	0.445	1.41 (0.58–3.38)	0.448	–	–
LDH > ULN	1.91 (0.00–4.32)	0.637	1.18 (0.59–2.39)	0.638	–	–
Primary melanoma subtype (other vs cutaneous)	1.64 (0.00–4.41)	0.521	1.31 (0.57–3.00)	0.523	–	–
ECOG performance score (poor vs good)	2.17 (0.00–5.13)	0.018	2.30 (1.13–4.63)	0.022	2.10 (1.05–4.46)	0.036
Clinical impression of LMD (yes vs no)	2.96 (1.23–4.69)	0.271	1.69 (0.66–4.39)	0.277	–	–
Prior systemic therapy (yes vs no)	2.96 (0.93–4.99)	0.899	1.06 (0.46–2.42)	0.899	–	–
Prior brain surgery (yes vs no)	1.54 (0.00–3.13)	0.035	2.05 (1.04–4.06)	0.039	1.89 (0.91–3.91)	0.086
LMD treatment (any vs no)	3.42 (1.77–5.07)	<0.001	0.13 (0.04–0.45)	0.001	0.16 (0.04–0.57)	0.005
LMD radiation (yes vs no)	3.55 (2.05–5.06)	0.015	0.43 (0.21–0.87)	0.019	–	–
Systemic LMD therapy (yes vs no)	3.06 (1.87–4.25)	0.559	0.80 (0.38–1.70)	0.561	–	–
Systemic LMD therapy only (yes vs no)	1.91 (0.00–3.96)	0.291	1.52 (0.69–3.34)	0.296
Systemic therapy while on steroids (yes vs no steroids at systemic therapy)	2.96 (1.33–4.59)	0.094	1.95 (0.88–4.31)	0.101	–	–
ICI therapy for LMD (yes vs no)	3.42 (2.19–4.65)	0.120	0.60 (0.31–1.16)	0.125	–	–
Targeted therapy for LMD (yes vs no)	4.34 (1.74–6.94)	0.476	0.79 (0.41–1.52)	0.478	–	–
Both radiation and systemic therapy for LMD (yes vs no)	3.42 (1.76–5.08)	0.197	0.65 (0.34–1.26)	0.202	–	–
Radiation only for LMD (yes vs no)	4.11 (0.65–7.57)	0.443	0.69 (0.27–1.78)	0.446	–	–
ICI therapy and radiation (yes vs no)	3.42 (0.00–9.94)	0.096	0.54 (0.26–1.13)	0.102	–	–
MEK/BRAF targeted therapy and radiation (yes vs no)	3.55 (1.77–5.33)	0.661	1.17 (0.58–2.38)	0.662	–	–
CSI (yes vs other radiation types)	5.44 (0.00–13.72)	0.492	0.65 (0.19–2.23)	0.495	–	–

Bolded values indicate statistical significance (P < 0.05).

BRAF, B-Raf proto-oncogene; CI, confidence interval; CNS, central nervous system; CSF, cerebrospinal fluid; CSI, craniospinal irradiation; ECOG, Eastern Cooperative Oncology Group; HR, hazard ratio; ICI, immune checkpoint inhibitor; KM, Kaplan–Meier; LDH, lactate dehydrogenase; LMD, leptomeningeal disease; MEK, mitogen-activated protein kinase, OS, overall survival; ULN, upper limit of normal.

Discussion

In our retrospective cohort of 44 melanoma patients with LMD, median OS was 3.0 months (Fig. 2a), consistent with prior studies reporting median OS ranging from 1.6 to 4.9 months [13,15,18–20]. Large retrospective series have similarly demonstrated poor survival: a single-institution study of 178 patients found a median OS of 3.5 months [13], a cohort of 110 patients reported 2.5 months [19], and a multicenter study of 172 melanoma patients with LMD reported OS of 4.9 months [15]. Notably, a recent nationwide Danish study analyzing patients with melanoma brain metastases reported outcomes of 67 patients with LMD having a much longer OS of 8.4 months [20]. This difference is due to the fact that patients with an LMD diagnosis in this cohort had to have LMD diagnosis done at the time of the discovery of intracranial metastases, and patients who developed LMD following treatment for intracranial disease, as seen in our cohort, were not included in the Danish study. This distinction underscores the persistently grim prognosis of melanoma-associated LMD among patients who experience progression in leptomeninges after multiple lines of treatment for metastatic melanoma. The median interval from diagnosis of metastatic disease in CNS to LMD diagnosis in our cohort was only 5.8 months (Table 1), highlighting the aggressive and rapidly evolving nature of LMD in pretreated melanoma patients.

Median PFS following the first systemic treatment initiation for LMD was 1.5 months (Table 1). Treatment strategies following LMD diagnosis were heterogeneous: 54.5% received both systemic therapy and radiation, 34.1% received monotherapy, and 11.4% received no treatment. Although 40.9% received ICIs and 43.2% received BRAF/MEK-targeted therapy, neither modality alone was significantly associated with a significant survival benefit in our analysis (Table 2). We observed a trend toward improved survival in patients receiving concurrent ICI and radiotherapy (hazard ratio 0.54; Fig. 2d), although this did not reach statistical significance, likely due to small sample size. On sensitivity analysis excluding a patient who died within 4 days of ICI initiation, survival differences became more pronounced, with median OS reaching significance by Kaplan–Meier estimation (4.1 months, P = 0.047) and the Cox model approaching significance (hazard ratio 0.47, P = 0.053; Fig. 2d, Table 2). Furthermore, 62.5% of patients treated with ICIs or BRAF/MEK inhibitors in our cohort received concurrent corticosteroids to control CNS symptoms that could blunt the efficacy of immune-based therapies [21]. In fact, patients who received ICIs or BRAF/MEK inhibitors with steroids had a nonsignificant trend toward worse survival (hazard ratio 1.95, P = 0.101; Table 2). Among radiation subtypes, craniospinal irradiation (CSI) did not confer a survival advantage over other approaches (hazard ratio 0.65, P = 0.495; n = 4), and this analysis is underpowered given only 4 patients of 31 received CSI.

In our cohort, recipients of both systemic and CNS-directed therapies did not have significant improvement in OS (Table 2), likely because 91% of patients in our cohort already progressed on prior systemic treatments and were receiving therapy with corticosteroids to control symptoms of CNS metastases.

Functional status was the strongest prognostic indicator in our study. Poor ECOG performance status (two or higher score) was independently associated with worse OS (hazard ratio 2.10) (Table 2, Fig. 2b), consistent with data from prior melanoma LMD cohorts [13,15,18]. Receipt of any treatment to control LMD was independently associated with improved OS (hazard ratio 0.16; Table 2, Fig. 2c), similar to other reports [2,13,18,19]. No significant associations were observed between OS and BRAF mutation status or elevated LDH (Table 2), factors previously reported as prognostic [13]. While neurological symptoms and elevated LDH have been linked to worse outcomes in prior cohorts, these associations did not reach significance in our dataset, potentially due to small sample size. Individuals with parenchymal brain metastasis showed a nonsignificant trend towards reduced survival, paralleling results from other LMD series [13].

CSF cytology to confirm imaging diagnosis of LMD was only done in only 9.1% of cases (Table 1). Molecular CSF diagnostics, such as the presence of tumor DNA, were not performed in our cohort. This is suggested in literature for confirmation and monitoring of LMD [7,22,23].

There is growing evidence [17] confirmed in our cohort that concurrent multimodal treatment, particularly ICIs and radiation, can offer benefit in LMD patients (Table 2, Fig. 2c and d). The frequent use of corticosteroids and high prevalence of LMD evolving in the context of progressing on ICI metastatic disease likely contribute to the modest results observed in our cohort as compared with other reported cohorts [1,11,24,25].

Novel therapies are needed for the treatment of LMD and recently reported data show the safety, feasibility, and activity of intrathecal immunotherapy [16,26,27]. CSF immune profiling has revealed that the presence of immunosuppressive dendritic and myeloid cell subpopulations correlate with worse prognosis. These may inform future therapeutic targets [4,6,25].

Conclusion

Melanoma-associated LMD remains a rapidly fatal despite the use of contemporary, multimodality therapies. In this multisite cohort, poor performance status was the strongest predictor of reduced survival, and the receipt of treatment for LMD was associated with improved outcomes. Combining ICI with radiation may provide benefits, although further investigation with a larger cohort is needed. These findings underscore the urgent need for prospective trials, improved diagnostics, and multimodal treatment strategies to improve outcomes in this vulnerable population.

Acknowledgements

We thank Shahin Kavousi, MD, for his contributions to data collection and validation, and Dr David M. Zahrieh, statistician, for statistical review.

This work was supported by Mayo Clinic Comprehensive Cancer Center Grant P30 CA015083.

Conflicts of interest

There are no conflicts of interest.