Integrative Analysis of Clinical Features and Cerebrospinal Fluid Immune Microenvironment in Leptomeningeal Metastasis From Non-small-cell Lung Cancer Patients

Abstract

Background

Leptomeningeal metastasis (LM) represents one of the most severe complications in advanced lung cancer, indicating a poor prognosis. However, the mechanisms underlying LM progression remain incompletely understood, and reliable biomarkers for monitoring disease progression and treatment response are lacking.

Methods

This retrospective observational study was conducted at the Zhejiang Provincial People’s Hospital. Cerebrospinal fluid (CSF) samples from 13 patients with non–small cell lung cancer (NSCLC) and LM were analyzed, including biochemical parameters, tumor markers, differential cell counts, and T-cell subsets assessed by flow cytometry.

Results

The median overall survival (mOS) after LM diagnosis was 15 months. During LM progression, increased CSF levels of albumin, lactate dehydrogenase (LDH), lactic acid, and carcinoembryonic antigen (CEA) were observed, accompanied by decreased glucose, chloride, lymphocyte proportion, and CD8⁺ T-cell proportion. In some cases, lymphocyte percentages and CD8⁺ T-cell proportions increased during subsequent evaluations following treatment adjustment.

Conclusions

These findings suggest that CSF biochemical and immune parameters may be associated with disease progression in LM. Dynamic changes in CD8⁺ T-cell proportions may also reflect alterations in the CSF immune microenvironment and could represent potential targets for future immunotherapeutic strategies.

Keywords

leptomeningeal metastasis lung cancer cerebrospinal fluid T-cell prognosis

Introduction

Leptomeningeal metastasis (LM), the infiltration of tumor cells into the cerebrospinal fluid (CSF)-filled leptomeninges, is a devastating complication of non-small-cell lung cancer (NSCLC), occurring in 3-5% of advanced patients, particularly with epidermal growth factor receptor (EGFR) mutations and anaplastic lymphoma kinase (ALK) rearrangements.^1-3 Despite multiple combined therapies including radiation, intrathecal treatment (IT), targeted therapy and immunotherapy, LM patients still have a poor prognosis with survival of less than 1 year.^4,5 The CSF, which bathes the central nervous system (CNS), plays a critical role in LM progression. Emerging evidence highlights the CSF immune microenvironment as a key determinant of metastatic colonization and therapeutic resistance, offering insights into the interplay between tumor cells and the CNS-specific immune landscape.

The CNS has historically been viewed as an immune-privileged site due to the blood-brain barrier (BBB). However, the CSF contains resident immune components, including macrophages, lymphocytes (predominantly CD4⁺ and CD8⁺ T cells), and cytokines, which collectively shape the local microenvironment.⁶ In LM, this balance is disrupted as metastatic NSCLC cells breach the BBB, colonize the leptomeninges, and interact with CSF immunocytes.⁷ Tumor cells exploit the unique immunological features of the CSF to evade detection and promote immunosuppression. Single-cell RNA sequencing of CSF in brain metastasis patients (62% lung cancer) revealed a heterogeneous immune infiltrate, including cytotoxic CD8⁺ T cells, regulatory T cells (Tregs), natural killer (NK) cells, and tumor-associated macrophages (TAMs).⁸ Nevertheless, the immune microenvironment in LM still remained elusive.

The CSF immune microenvironment offers actionable insights. CSF ctDNA outperforms plasma in detecting LM-specific mutations, enabling early diagnosis and monitoring.⁹ Immune profiling via CSF-derived exosomes or T cell clonotypes could predict responses to immunotherapy, such as immune checkpoint inhibitors (ICIs), which show promise in LM patients with high CD8⁺ T cell infiltration.¹⁰

However, the CSF immune microenvironment in NSCLC-LM is still a complex ecosystem shaped by tumor-driven remodeling, genetic evolution, and immune evasion that is poorly understood. In this study, we collected CSF from NSCLC patients with LM who received the treatment combination of Ommaya reservoir implantation (ORI) and ventriculoperitoneal shunt (VPS) and followed by IT, and analysis the clinical features and CSF immune micro-environment to explore the mechanism of onset and development of LM.

Method

Study Design and Patient Population

We conducted a retrospective observational study including 13 patients with non-small cell lung cancer (NSCLC) and leptomeningeal metastasis (LM) treated at Zhejiang Provincial People’s Hospital between January 1, 2020 and August 31, 2025. All patients were consecutively enrolled, and all clinical data were de-identified prior to analysis. No sample size calculation was performed due to the retrospective nature of the study. The reporting of this study conforms to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.¹¹

The exclusion criteria were as follows: patients who were unable to tolerate or cooperate with intrathecal therapy; patients with concurrent systemic malignancies other than lung cancer; patients with craniocerebral trauma or intracranial infection; patients with autoimmune diseases; and patients with pre-existing neurological disorders.

LM was confirmed by positive CSF cytology together with clinical and neuroimaging findings. All patients underwent VPS placement and ORI which enabled CSF diversion for intracranial pressure control and facilitated repeated intrathecal drug administration and CSF sampling. The Ommaya reservoir is a small implantable device placed beneath the scalp and connected to the lateral ventricle via a catheter, allowing repeated administration of chemotherapeutic agents and CSF sampling without the need for lumbar puncture.¹² VPS is commonly used for CSF diversion by draining CSF from the cerebral ventricles to the peritoneal cavity to relieve intracranial hypertension.¹³

In our clinical practice, a standard programmable VPS system and an Ommaya reservoir (RE-2021, SOPHYSA) were implanted according to neurosurgical indications. The Ommaya reservoir was used for intrathecal drug administration and CSF sampling. During intrathecal drug delivery,the duration of valve adjustment was determined based on clinical judgment and individual patient condition. Due to the retrospective nature of this study, the exact management protocol was not fully standardized across all patients.

Clinical data were collected from medical records, including sex, age, pathological diagnosis, genetic status, LM-related symptoms, metastatic status, and survival outcomes.

Treatment response was evaluated after every two cycles of intrathecal therapy according to the Response Assessment in Neuro-Oncology Leptomeningeal Metastases (RANO-LM) criteria. The assessment integrated neurological examination, neuroimaging findings, and cerebrospinal fluid (CSF) cytology results. According to the RANO-LM framework, disease status is determined using a composite evaluation of clinical neurological function, CSF tumor cell detection, and contrast-enhanced MRI of the neuraxis. In the present study, patients were categorized into stable disease (SD) and progressive disease (PD) status. Because objective radiographic or cytological response is uncommon in leptomeningeal metastasis and stabilization is often regarded as the most clinically meaningful therapeutic outcome, patients demonstrating response or SD were analyzed together as the SD status while patients showing clinical, radiographic, or cytological progression were classified as the PD status.¹⁴ For comparison of immune profiles, CSF samples were also collected from 5 patients with benign brain tumors who underwent neurosurgical procedures during the same period.

This study was approved by the Institutional Review Board (IRB) and Ethics Committee of Zhejiang Provincial People’s Hospital (Approval No. 2025-140; Date of approval: June 9, 2025) and was conducted in accordance with the Declaration of Helsinki. The requirement for informed consent was waived by the IRB due to the retrospective design of the study.

Intrathecal Chemotherapy

Intrathecal chemotherapy was administered through the Ommaya reservoir according to institutional clinical practice. The regimens included pemetrexed (PEM), thiotepa (TSPA), and methotrexate (MTX). PEM was administered at a dose of 20 mg combined with 5 mg dexamethasone (DXMS) on Days 1 and 8 of a 21-day cycle. TSPA was administered at 10 mg combined with 5 mg DXMS on Days 1, 8, and 15 of a 28-day cycle. MTX was administered at 10 mg combined with 3 mg DXMS on Days 1 and 5 of a 21-day cycle.All intrathecal drugs were delivered via the Ommaya reservoir to allow repeated administration and standardized CSF sampling during treatment.

CSF Sample Testing

5 mL of CSF was collected prior to each cycle of intrathecal chemotherapy. Routine CSF examinations included pressure measurement, total cell counts, differential cell counts, biochemical analysis, and carcinoembryonic antigen (CEA) testing. Differential cell counts were determined manually by microscopy. Biochemical parameters, including glucose, chloride, albumin, lactate dehydrogenase (LDH), and lactic acid, were measured using an automated biochemical analyzer (Beckman Coulter AU5800, USA) with corresponding AU Chemistry Systems reagents. CSF CEA levels were determined using a chemiluminescent immunoassay on the Abbott Alinity i system (Abbott Laboratories, USA) with the Alinity i CEA assay kit.CSF cytology was evaluated after 2–3 cycles of intrathecal therapy to assess the presence of malignant cells.

Immunotyping by Flow Cytometry

10 mL of CSF was collected prior to each cycle of intrathecal chemotherapy for flow cytometry analysis. In addition, 10 mL of CSF from 5 patients with benign brain tumors undergoing neurosurgical procedures during the same period was collected as the control group.

CSF samples were processed within 2 hours of collection. No viability dye was used during the analysis.CSF-derived cells were incubated with fluorescence-labeled monoclonal antibodies against CD3, CD45, CD4, and CD8 under dark conditions on ice according to the manufacturer’s protocol. The antibodies used included mouse anti-human CD3-FITC (UCHT1, UB104411), CD45-PerCP-Cy5.5 (HI30, UB109481), CD4-PC7 (SK3, UB105441), and CD8-APC (SK1, UB106421) (UBBio Technology, Zhejiang, China).Flow cytometry acquisition was performed using a NovoCyte flow cytometer (Agilent Technologies, Santa Clara, CA, USA) and analyzed with NovoExpress software (version 1.4.1). During analysis, debris and doublets were excluded using forward- and side-scatter gating, and T-lymphocyte subsets were quantified after gating on CD3⁺ T cells. The proportions of CD4⁺ and CD8⁺ T cells were expressed as percentages within the CD3⁺ T-cell population.

Statistical Analysis

Because of the retrospective design, not all measurements were available for every patient at each disease state, and sample sizes therefore varied across analyses. Statistical analyses were performed using SPSS version 26.0 (IBM Corp., Chicago, IL, USA). Continuous variables were assessed for distribution before analysis and are presented as mean ± standard error of the mean (SEM) when normally distributed. Between-group comparisons were performed using Student’s t-test for normally distributed variables and the Wilcoxon rank-sum test for non-normally distributed variables. For disease-state comparisons, all available observations at the corresponding status were included, with the number of observations reported in the figure legends.Overall survival (OS) was defined as the time from LM diagnosis to death from any cause. Survival was estimated using the Kaplan–Meier method, and patients who were alive at the end of follow-up (August 31, 2025) were censored at their last follow-up. All tests were two-sided. Exact P values were reported to four decimal places where applicable, and values <0.0001 were reported as P < 0.0001.

Results

The Characteristics of NSCLC Patients With LM

This observational study included 13 patients with confirmed lung adenocarcinoma, whose demographic and clinical characteristics were summarized in Table 1. EGFR mutations were identified in 76.9% of cases (n=10/13). At baseline, 15.4% of participants (n=2/13) presented with LM, while 46.2% (n=6/13) exhibited bone metastases, one case with concurrent LM and 5 cases developing LM post-PD. The principal clinical manifestations of LM comprised of headache, nausea, and vomiting. CSF pressure elevation was universally observed upon LM diagnosis. All patients received concurrent systemic therapy in addition to intrathecal chemotherapy throughout the treatment cycles. At the end of follow-up (August 31, 2025), the median OS was 15 months (Figure 1). 5 of 13 patients (38.5%) remained alive and were censored at their last follow-up, with the longest recorded survival period exceeding four years following initial diagnosis. It should be noted that this cohort represents a highly selected subgroup of LM patients who were eligible for VPS and intrathecal therapy, which may influence the interpretation of overall survival outcomes.

Table 1.

The Patients’ Characteristics From Baseline to LM

No.	Gender	Age	Pathology	Gene status	PD-L1 (%)	Stage in baseline	Metastasis in baseline	Metastasis in PD	Symptom of LM	*CSF pressure (mmH₂O)	Time from baseline to LM (Mo)	Cycles of IT	^#Survival (Mo)	OS	Cause of death
1	M	68	LU	negat-ive	10	IVB	LM	same as baseline	dizziness,memory loss	220	0	30	39	censored	/
1	M	68	AD	negat-ive	10	IVB	LM	same as baseline	dizziness,memory loss	220	0	30	39	censored	/
2	F	58	LU	EGFR L861Q	/	IVB	bone, Liver, brain,LM	same as baseline	epilepsy, headache	>400	0	1	2	2	hemorrhage
2	F	58	AD	EGFR L861Q	/	IVB	bone, Liver, brain,LM	same as baseline	epilepsy, headache	>400	0	1	2	2	hemorrhage
3	M	62	LU	EGFR L858R	1	IVB	bone	LM	dizziness, tinnitus	>400	28	28	35	censored	/
3	M	62	AD	EGFR L858R	1	IVB	bone	LM	dizziness, tinnitus	>400	28	28	35	censored	/
4	F	78	LU	EGFR L858R	/	IIIA	no	brain,intra-pulmonary	nausea, vomit	225	72	2	4	4	septicemia
4	F	78	AD	EGFR L858R	/	IIIA	no	brain,intra-pulmonary	nausea, vomit	225	72	2	4	4	septicemia
5	M	63	LU	EGFR 19DEL	/	IIIA	no	LM	headache,unsteady gait	>400	62	40	58	censored	/
5	M	63	AD	EGFR 19DEL	/	IIIA	no	LM	headache,unsteady gait	>400	62	40	58	censored	/
6	F	51	LU	EGFR L861Q	/	IVB	bone, intrapul-monary	LM	nausea, vomit	360	11	8	12	12	progress of LM
6	F	51	AD	EGFR L861Q	/	IVB	bone, intrapul-monary	LM	nausea, vomit	360	11	8	12	12	progress of LM
7	F	55	LU	EGFR L858R	/	IIIA	no	brain,intra-pulmonary, bone,LM	headache,nausea, vomit	310	61	11	15	15	progress of LM
7	F	55	AD	EGFR L858R	/	IIIA	no	brain,intra-pulmonary, bone,LM	headache,nausea, vomit	310	61	11	15	15	progress of LM
8	M	53	LU	KRAS G12D	/	IVB	liver	brain,LM,liver	muscle weakness,sphasia	190	26	4	3	censored	/
8	M	53	AD	KRAS G12D	/	IVB	liver	brain,LM,liver	muscle weakness,sphasia	190	26	4	3	censored	/
9	M	47	LU	EGFR 19DEL	/	IVB	bone, intrapul-monary	LM	dizziness,nausea	>400	13	14	26	26	intracranial infection
9	M	47	AD	EGFR 19DEL	/	IVB	bone, intrapul-monary	LM	dizziness,nausea	>400	13	14	26	26	intracranial infection
10	M	51	LU	EGFR 19DEL	/	IIIA	no	LM,brain	dizziness,nausea, vomit	>400	32	22	28	censored	/
10	M	51	AD	EGFR 19DEL	/	IIIA	no	LM,brain	dizziness,nausea, vomit	>400	32	22	28	censored	/
11	F	62	LU	EGFR L858R	/	IVB	bone	LM,intrapulmonary	nausea, vomit, epilepsy	300	48	4	7	7	progress of LM
11	F	62	AD	EGFR L858R	/	IVB	bone	LM,intrapulmonary	nausea, vomit, epilepsy	300	48	4	7	7	progress of LM
12	M	83	LU	EGFR L858R	/	IIA	no	LM,brain,intrapulmonary	epilepsy	190	132	2	6	6	respiratory failure
12	M	83	AD	EGFR L858R	/	IIA	no	LM,brain,intrapulmonary	epilepsy	190	132	2	6	6	respiratory failure
13	M	67	LU	negat-ive	/	IVB	bone	LM,brain	headache	110	11	2	4	4	progress of LM
13	M	67	AD	negat-ive	/	IVB	bone	LM,brain	headache	110	11	2	4	4	progress of LM

*CSF pressure at initial diagnosis of LM.

#Survival up to now after LM.

Figure 1.

Kaplan–Meier survival curve showing overall survival (OS) of 13 patients with NSCLC after the diagnosis of LM. Tick marks indicate censored observations. OS was defined as the time from LM diagnosis to death from any cause. Patients alive at the end of follow-up were censored.This cohort represents a highly selected subgroup of patients eligible for VPS and intrathecal therapy, which may influence the interpretation of survival outcomes

Correlation Between Biochemical Index of CSF and the Efficacy of Patients’ Treatment

We compared biochemical indexes of CSF in PD and SD status in 13 NSCLC patients with LM respectively. The level of albumin in the CSF of PD status was higher than SD status (Figure 2A). The levels of LDH (Figure 2D) and lactic acid (Figure 2E) were also increased when LM progressed. Conversely, the levels of glucose (Figure 2B) and chloride (Figure 2C) in the cerebrospinal fluid decreased when the patients were in PD status.

Figure 2.

The biochemical index of CSF in NSCLC patients with LM. The levels of albumin (A), glucose (B), chloride (C), lactate dehydrogenase (LDH) (D), and lactic acid (E) were compared between LM-SD status and LM-PD status. The number of observations included in each comparison was as follows: albumin, n = 72 (SD) and n = 47 (PD); glucose, n = 74 (SD) and n = 50 (PD); chloride, n = 63 (SD) and n = 44 (PD); LDH, n = 67 (SD) and n = 45(PD); lactic acid, n = 68 (SD) and n = 44(PD)

Correlation Between CSF Tumor Biomarker and the Efficacy of Patients’ Treatment

Here, given that all the patients with LM were diagnosed with lung adenocarcinoma (LUAD), we selected carcinoembryonic antigen (CEA) as the tumor biomarker to predict PD in these patients. Consistent with our hypothesis, CEA levels in the CSF were markedly elevated during PD (Figure 3A). However, no significant difference in serum CEA levels was observed between the SD status and PD status (Figure 3B).

Figure 3.

CSF tumor biomarker in NSCLC patients with LM. The levels of CEA in the CSF (A) and serum (B) were compared between SD status and PD status. The number of observations included in each comparison was as follows: CEA in CSF, n=13 (SD) and n=15(PD); CEA in serum, n=9 (SD) and n=11(PD)

Association Between Dynamic Changes in CSF Differential Cell Counts and Therapeutic Efficacy in LM Patients

We performed differential cell counts in the CSF, with a focus on the proportions of lymphocytes and macrophages. During follow-up, a gradual decline in the proportion of lymphocytes (Figure 4A) and a progressive increase in macrophages (Figure 4B) within the CSF were observed with disease progression. Notably, the lymphocyte proportion rebounded following the initiation of adjusted treatment that elicited an effective therapeutic response.

Figure 4.

The differential cell counts in NSCLC patients with LM. The proportions of lymphocytes and macrophages were compared between SD status (n=49) and PD status (n=37)

Dynamic Changes in T-Lymphocyte Subsets in NSCLC Patients With LM

We further analyzed CSF T-lymphocyte subsets by flow cytometry in patients with benign brain tumors (control group) and in LM patients during LM-SD and progressive disease LM-PD status. The control group exhibited a significantly higher proportion of CD8⁺ T cells compared with both LM-SD and LM-PD status (Figure 5B). In contrast, no significant differences were observed in CD4⁺ T-cell proportions or the CD4⁺/CD8⁺ T-cell ratio between the control and LM-SD status (Figure 5A,C).During disease progression from LM-SD to LM-PD, a progressive decrease in CD8⁺ T-cell proportions was observed, accompanied by increases in CD4⁺ T-cell proportions and the CD4⁺/CD8⁺ T-cell ratio (Figure 5A-C). In several patients who achieved clinical stabilization following treatment adjustment, CD8⁺ T-cell proportions increased during subsequent evaluations. However, no viability dye was used during the analysis, which should be considered when interpreting these results.

Figure 5.

Dynamic changes in T-lymphocyte subsets in NSCLC patients with leptomeningeal metastasis (LM).The proportions of CD4⁺ T cells (A), CD8⁺ T cells (B), and the CD4⁺/CD8⁺ T-cell ratio (C) were compared among the control group, LM-SD status, and LM-PD status. T-cell proportions were calculated within the CD3⁺ T-cell population. The number of observations included in each comparison was shown: n = 5 (control), n = 17 (SD), and n = 15(PD). (D) Representative flow cytometry plots from a control patient and an LM patient during SD and PD status. No viability dye was used during the analysis. aControl vs. LM LMSD, bControl vs. LM-PD,cLM-SD vs. LM-PD

Discussion

LM is an aggressive manifestation marked by the spread of tumor cells into the CSF, subarachnoid space, and leptomeninges.¹⁵ Recent advances have deepened insights into leptomeningeal pathology, revealing a distinct and varied immune landscape in this compartment compared to systemic, brain parenchymal, or dural immune microenvironments. Notably, previous studies highlight dysregulated immune responses during tumor progression, rendering LM challenging to manage.¹⁶ Single-cell RNA sequencing further reveals CSF macrophages adopting immunosuppressive, pro-tumorigenic M2 polarization.¹⁷ Additionally, CSF in LM patients shows increased proportions of exhausted T cells and regulatory T cells relative to controls, suggesting an immune environment permissive to tumor survival.¹⁸ These findings show the complexity of immune-tumor interactions in LM, while there remain substantial unknowns requiring elucidation.

In this study, we analyzed a retrospective cohort of 13 patients with LUAD who developed LM to explore the relationship between clinical characteristics and the CSF immune microenvironment. LM was identified in 15.4% of patients at baseline and became more frequent after disease progression, suggesting that LM is often a complication of advanced-stage disease. The most common symptoms observed in LM patients, including headache, nausea, and vomiting, are likely associated with increased intracranial pressure caused by CSF circulation disturbances. Despite the presence of metastatic disease, the median OS in this cohort was 15 months. At the follow-up cutoff, 38.5% of patients were still alive, including one patient with survival exceeding four years. As some patients had not yet reached the study endpoint at the time of analysis and were treated as censored observations, the survival estimates should be interpreted cautiously. Longer follow-up may further refine the survival profile of this selected clinical population. Bander et al.’s study showed that CSF diversion for LM with hydrocephalus and intracranial hypertension secondary to metastasis can achieve symptomatic relief, hospital discharge, and return to further oncologic therapy.¹⁹ Similarly, in our cohort, the use of VPS and ORI facilitated both symptomatic control and repeated intrathecal drug administration, which likely contributed to improved clinical management and enabled long-term follow-up of these patients. The obvious elevation of CSF albumin in the PD status compared to SD status likely reflects increased BBB disruption during LM progression, a phenomenon linked to tumor infiltration, inflammation, or vascular leakage.²⁰ Higher lactate dehydrogenase (LDH) and lactic acid levels in PD further support tumor active status, as LDH is a complex biomarker associated with the activation of several oncogenic signaling pathways as well as with the metabolic activity, invasiveness and immunogenicity of cancer.²¹ High LDH in the CSF was considered as a poor prognostic indicator in neoplastic meningitis.²² While lactic acid accumulation may indicate anaerobic metabolism driven by hypoxia in rapidly proliferating metastatic niches.²³ In contrast, reduced CSF glucose and chloride levels observed in our cohort may be explained by increased metabolic consumption by proliferating tumor cells in the CSF compartment.^24,25 Together, these biochemical alterations are consistent with the pathophysiological changes associated with LM progression and disruption of CSF homeostasis.

We also focused on the differential utility of CEA in CSF versus serum for predicting PD in LUAD patients with LM. Elevated CSF CEA during PD hinted that local biomarker levels better reflect CNS-specific progression. In contrast, serum CEA level was lack of correlation may stem from systemic influences, diluting its specificity for LM progression. These observations highlight the importance of compartment-specific biomarkers when evaluating metastatic disease involving the central nervous system.

Meanwhile, we observed a shift in CSF cellular dynamics, characterized by decreased lymphocyte proportions and increased macrophage proportions during PD, which may reflect changes in the immune microenvironment associated with LM progression. Reduced lymphocyte levels may indicate impaired anti-tumor immunity, while increased macrophages may represent tumor-associated macrophages (TAMs) contributing to an immunosuppressive microenvironment.^26,27 The increase of lymphocytes indicated restored immune activity during therapeutic response, suggesting lymphocyte proportions may correlate with treatment efficacy. Analysis of T-lymphocyte subsets further showed that LM patients had lower CD8⁺ T-cell proportions (both SD and PD status) compared with benign controls, suggesting compromised cytotoxic immune activity in the CSF compartment. Moreover, during progression from SD to PD, CD8⁺ T-cell proportions decreased, whereas CD4⁺ T-cell proportions and the CD4⁺/CD8⁺ ratio increased, indicating a potential shift toward an immunosuppressive T-cell profile. These findings are consistent with previous studies. Smalley et al found the CSF of melanoma patients with LM showed the highest number of apoptotic and exhausted CD4⁺ T-cells and the lowest number of CD8⁺ T-cells, indicating the existence of an immune-suppressed T cell microenvironment.²⁸ Ruan et al’ study showed that compared with control samples, CSF samples from breast cancer and lung cancer patients with LM had naïve CD4⁺ T-cells, exhausted and cytotoxic CD8⁺ T-cells, a slightly higher number of Treg and alternatively activated macrophages.¹⁸ Latest research indicated that PD-1–positive T cells are enriched in the tumor microenvironment of leptomeningeal metastases compared with primary lung tumors in murine models of lung-to-leptomeningeal metastasis.²⁹ Effective anti -tumor treatment was accompanied by the restoration of CD8⁺ T-cells, suggesting that CD8⁺ T-cells may serve as potential immunotherapeutic targets for LM.

Several limitations should be acknowledged. This was a retrospective study with a relatively small sample size, which may limit statistical power and generalizability. In addition, the cohort represents a selected subgroup of LM patients who were able to undergo VPS and intrathecal therapy, and therefore the survival findings should be interpreted with caution. Some patients were still alive at the end of follow-up, which may affect the estimation of overall survival.There was also heterogeneity in systemic and intrathecal treatments across patients, which could influence CSF biochemical and immune parameters. Furthermore, no viability dye was used during flow cytometry analysis, which may affect the accuracy of immune cell quantification. This should be considered when interpreting the results.Future studies with larger cohorts and more standardized protocols are needed to validate these findings and better define the role of CSF immune profiling in leptomeningeal metastasis.

Conclusion

Our study suggests that in NSCLC patients with LM, several CSF biomarkers, including albumin, glucose, chloride, LDH, lactate, CEA, lymphocyte proportion, and CD8⁺ T-cell proportion, may be associated with disease progression. These findings highlight the potential value of CSF immune and biochemical profiling in monitoring LM. Further studies with larger cohorts are needed to better characterize the CSF tumor immune microenvironment and to develop multivariable predictive models for LM.

Footnotes

Acknowledgements

The authors thank all participants and patients involved in this study.

ORCID iD

Yun Chen

Ethical Considerations

This study was approved by the Institutional Review Board (IRB) and Ethics Committee of Zhejiang Provincial People’s Hospital (Approval No. 2025-140; Date of approval: June 9, 2025). The approval covered the study population and follow-up period included in the present analysis, including follow-up through August 31, 2025. The study was conducted in accordance with the principles of the Declaration of Helsinki.

Consent to Participate

The requirement for informed consent was waived by the IRB due to the retrospective nature of the study and the use of de-identifieddata.

Author Contributions

Junjun Chen: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Writing-original draft; Writing -review & editing. Qihao Zhou: Data curation; Supervision. Yuan Yin: Data curation; Investigation; Zhiquan Qin: Investigation;Writing-review & editing. Yun Chen: Conceptualization; Funding acquisition; Supervision; Writing-review & editing.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by Zhejiang Provincial Medical and Health Science and Technology Programme (grant numbers 2024KY649) and Key Discipline of Traditional Chinese Medicine in Zhejiang Province: Integrated Clinical Medicine (Oncology) (grant numbers 2024-XK-03).

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The datasets used and/or analyzed during this study are available from the corresponding author on reasonable request.*

Use of Artificial Intelligence statement

Generative AI tools were used only for language polishing and formatting assistance. No AI tools were used to generate or modify scientific data, perform statistical analysis, create research results, or draw scientific conclusions. All content was reviewed and approved by the authors, who take full responsibility for the final manuscript.

References

Remsik

Boire

. The path to leptomeningeal metastasis. Nat Rev Cancer. 2024;24:448-460.

Ozcan

Singh

Vredenburgh

. Leptomeningeal Metastasis from Non-Small Cell Lung Cancer and Current Landscape of Treatments. Clin Cancer Res. 2023;29:11-29.

Murciano-Goroff

Harada

Drilon

. An Ascendant Challenge: Central Nervous System Metastases in ALK+ Lung Cancers. Clin Cancer Res. 2022;28:2477-2479.

Oyoshi

Hirata

Hirano

, et al. Prognostic impact of EGFR/ALK alterations in leptomeningeal metastasis from lung adenocarcinoma treated with whole-brain radiotherapy. Clin Exp Metastasis. 2023;40:407-413.

Le Rhun

Devos

Weller

, et al. Prognostic validation and clinical implications of the EANO ESMO classification of leptomeningeal metastasis from solid tumors. Neuro Oncol. 2021;23:1100-1112.

Piehl

van Olst

Ramakrishnan

, et al. Cerebrospinal fluid immune dysregulation during healthy brain aging and cognitive impairment. Cell. 2022;185:5028-5039.

Alexander

Lin

Cheng

. Leptomeningeal Metastases in Non-small Cell Lung Cancer: Optimal Systemic Management in NSCLC With and Without Driver Mutations. Curr Treat Options Oncol. 2020;21:72.

Rubio-Perez

Planas-Rigol

Trincado

, et al. Immune cell profiling of the cerebrospinal fluid enables the characterization of the brain metastasis microenvironment. Nat Commun. 2021;12:1503.

Zheng

Zhou

Chen

, et al. Cerebrospinal fluid circulating tumor DNA profiling for risk stratification and matched treatment of central nervous system metastases. Nat Med. 2025;31(5):1547-1556.

10.

Huang

, et al. Serum Exosomal miRNA Might Be a Novel Liquid Biopsy to Identify Leptomeningeal Metastasis in Non-Small Cell Lung Cancer. Onco Targets Ther. 2021;14:2327-2335.

11.

von Elm

Altman

Egger

Pocock

Gøtzsche

Vandenbroucke

. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Ann Intern Med. 2007;147(8):573-577.

12.

Hall

. Orlando De Jesus.Ommaya Reservoir. In: StatPearls Treasure Island (FL). StatPearls Publishing; 2025.

13.

Oliveira

Hakim

Semione

GDS

, et al. Ventriculoatrial Shunt Versus Ventriculoperitoneal Shunt: A Systematic Review and Meta-Analysis. Neurosurgery. 2024:903-915.

14.

Chamberlain

Junck

Brandsma

, et al. Leptomeningeal metastases: a RANO proposal for response criteria. Neuro Oncol. 2017;19(4):484-492.

15.

Sharma

Low

Kumthekar

. Advances in the Diagnosis and Treatment of Leptomeningeal Disease. Curr Neurol Neurosci Rep. 2022;22:413-425.

16.

Rua

McGavern

. Advances in Meningeal Immunity. Trends Mol Med. 2018;24:542-559.

17.

Remsik

Tong

Kunes

, et al. Leptomeningeal anti-tumor immunity follows unique signaling principles. bioRxiv, 2023,2023.03.17.533041.

18.

Ruan

Wang

Sun

, et al. Single-cell RNA sequencing reveals the characteristics of cerebrospinal fluid tumour environment in breast cancer and lung cancer leptomeningeal metastases. Clin Transl Med. 2022;12:e885.

19.

Bander

Yuan

Reiner

, et al. Cerebrospinal fluid diversion for leptomeningeal metastasis: palliative, procedural and oncologic outcomes. J Neurooncol. 2021;154:301-313.

20.

Jiang

Wang

Sun

. Novel method for the detection and quantification of malignant cells in the CSF of patients with leptomeningeal metastasis of lung cancer. Oncol Lett. 2016;11:619-623.

21.

Claps

Faouzi

Quidville

, et al. The multiple roles of LDH in cancer. Nat Rev Clin Oncol. 2022;19:749-762.

22.

Cacho-Díaz

Lorenzana-Mendoza

Reyes-Soto

, et al. Lactate dehydrogenase as a prognostic marker in neoplastic meningitis. J Clin Neurosci. 2018;51:39-42.

23.

Nakamizo

Sasayama

Shinohara

, et al. GC/MS-based metabolomic analysis of cerebrospinal fluid (CSF) from glioma patients. J Neurooncol. 2013;113(1):65-74.

24.

Duan

Lei

, et al. Utility and performance of cerebrospinal fluid cytology in discriminating central nervous system infections and brain tumors. J Neurooncol. 2025;175(3):1343-1353.

25.

Xie

Kang

Huang

. Intrathecal methotrexate injection combined treatment of gastric cancer with leptomeningeal metastasis: case report. Front Oncol. 2024;14:1464376.

26.

Räuber

Schulte-Mecklenbeck

Willison

, et al. Flow cytometry identifies changes in peripheral and intrathecal lymphocyte patterns in CNS autoimmune disorders and primary CNS malignancies. J Neuroinflammation. 2024;21:286.

27.

Hori

Sasayama

Tanaka

, et al. Tumor-associated macrophage related interleukin-6 in cerebrospinal fluid as a prognostic marker for glioblastoma. J Clin Neurosci. 2019;68:281-289.

28.

Smalley

Chen

Phadke

, et al. Single-Cell Characterization of the Immune Microenvironment of Melanoma Brain and Leptomeningeal Metastases. Clin Cancer Res. 2021;27:4109-4125.

29.

Kanaya

Seddiq

Chen

, et al. Engineered allogeneic stem cells orchestrate T lymphocyte-driven immunotherapy in immunosuppressive leptomeningeal brain metastasis. J Natl Cancer Inst. 2025;117:1151-1165.