Abstract Leptomeningeal disease (LMD) is one of the most devastating complications of advanced cancer, resulting in severe neurological deterioration and poor survival outcomes. Despite this, very little is known about the mechanisms responsible for the exceedingly poor prognosis and severely impaired neurological function of these patients. Here, we report that toxic branched-chain keto acids (BCKAs) accumulate in the cerebrospinal fluid of LMD patients and drive both immunosuppression and neurodegeneration, presenting a novel therapeutic target. Using single-cell transcriptomics and multi-omics profiling of patient specimens, we created a comprehensive LMD atlas that revealed two hallmarks of LMD: an immunosuppressive cellular microenvironment and an environment promoting neurodegeneration. Mechanistically, there is a profound BCKA-mediated disruption of both T cell function and neuronal integrity. BCKA exposure inhibits T cell viability, proliferation, energy metabolism, and effector cytokine release. Strikingly, BCKA-reducing therapy using sodium phenylbutyrate improves survival and neurological outcomes in preclinical models while enhancing the efficacy of CAR-T cell therapy in lymphoma LMD and immunecheckpoint inhibitors in melanoma LMD. Thus, BCKA accumulation is a central mediator of LMD progression and treatment resistance and provides an actionable therapeutic opportunity through drug repurposing. Importantly, our pan-cancer atlas of human LMD delivers a framework for understanding disease mechanisms and developing novel therapeutic strategies for this urgent unmet clinical need.

Leptomeningeal disease (LMD) occurs when malignant cells seed into the leptomeningeal space and cerebrospinal fluid (CSF), leading to severe neurological symptoms and very short survivals. LMD occurs in 5to15% of cancer patients and is most common in breast, melanoma, lung, and lymphomas. LMD tumors are generally resistant to all standard of care therapies. To better understand LMD biology we used comprehensive multiomic analyses of CSF specimens from LMD patients with different primary tumors to understand its immune and metabolic tumor microenvironment. Single cell RNA sequencing on patient CSF showed a distinct immunosuppressive landscape regardless of the primary histology. We identified significant enrichment of CD4Tregs and exhausted T cells in LMD patients. Lack of active, proliferating T cells and protumorigenic macrophage infiltration were associated with LMD poor survival (<10months, p<0.01). Additionally, proteomic and lipidomic analysis of patients’ CSF showed downregulation in proteins and lipids vital for neuronal development and myelin sheath integrity. Strikingly, the metabolomic analysis demonstrated an accumulation of branched-chain keto acids (BCKA), well known neurotoxins, in CSF of LMD versus controls (p<0.05). Next, we performed correlation analysis between single cell RNA seq data and levels of BCKA in LMD samples (n=8). Results revealed positive correlations between high BCKA concentrations and CD4Tregs (p<0.01), and two exhausted T cell populations (p<0.05), suggesting an immunosuppressive impact of BCKA in LMD patients. Our in-vitro functional data demonstrated that BCKA drastically disrupts the T-lymphocyte viability, proliferation, and effector cytokine secretion. Moreover, BCKA reduced the viability of chimeric antigen receptor CART cells, neurons, and meningeal cells but not tumor cells. These results were validated in immunocompetent LMD animal models. Consistent with BCKA-induced neurotoxicity, we found that LMD mice had a rapid neurological decline (p<0.0001). Furthermore, we found the leptomeningeal layer in LMD mice was compromised and had significantly high levels of BCKA (p<0.01). Highlighting the translatability of this work, we found that phenylbutyrate, a BCKA lowering agent, improved neurological function (p<0.01), survival outcomes (p<0.01), and efficacy of CART therapy in lymphoma LMD model (p<0.05). Phenylbutyrate also improved responses to chemotherapy in breast cancer LMD model. This is the first report describing BCKA accumulation in LMD as a cause of neurotoxicity and immunosuppression and provides a unique strategy to treat these deadly tumors. Repurposing of FDA approved phenylbutyrate improves the quality of life, survival, and efficacy of immune therapies in LMD. Mariam Lotfy Khaled, Ethan Vallebuona, Min Liu, M.Baraa Boozo, Zhihua Chen, Gerald C. Wallace, Yuan Ren, Ronak Kundalia, Hasan Alhaddad, Oscar Ospina, Brittany Evernden, Victoria Izumi, Lancia NF Darville, Ann chen, MacLean Hall, Michael Jain, Shari Pilon-Thomas, Paul Stewart, Fredrick L. Locke, Timothy J. Robinson, John M. Koomen, Peter A. Forsyth, Inna Smalley. Accumulation of branched chain keto acids promotes immunosuppression and neurodegeneration in leptomeningeal disease [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 3804.

The CSF of patients with LMD have an innate, but not adaptive, immune cellular profile. Murine LMD models showed IT cDC1s were safe, induced a Th1 response (CD4+ > CD8+ T cell and B cell dependent), cured most HER2+ LMD, and prevented LMD recurrence. We conducted a Phase I study of IT DCs and hypothesized the CSF would be remodeled to have a Th1 adaptive immunological profile. This is a Phase I single-arm, dose escalation study to establish 1) the safety and 2) associations between clinical outcomes & translational CSF studies. Eligibility includes TNBC or HER+ LMD pts, prior pCSpRT/WBRT, ECOG PS ≤2, and Ommaya reservoir. IT cDC1 IT were administered weekly x 12 at one of 4 dose levels (1 x 106 – 5 x 107) until PD/DLT. Endpoints were 1) safety, MTD RP2D, 2) association between clinical and immune profiles in the CSF. DLTs were defined as ≥ gr. 3 not due to LMD & didn’t respond to intervention. As of 03/15/25, the first 6 patients received 1 x 106 – 2 x 107 DCs with no DLTs. Headaches were common and grade 3 in two (33%) patients. Two (33%) had PD of LMD within 13 weeks. One patient died from leukoencephalopathy. The median OS was 48.0 weeks and five (83%) are alive. CSF cytokines Th1 cytokines included IFN-g, IL-6, IL-12, TNF-a; Th2-related IL-4 was not elevated. Transcriptomics showed a marked increase in CD4+ T, CD8+ T and B cells with a reduction of tumor cells. Antibodies to HER2 or HER3 developed in four of five (80%) patients. Our results suggest this approach activates CD4+ Th1 adaptive immune responses that drives adaptive antitumor activity which is otherwise lacking in LMD (NCT05809752).

BACKGROUND:Leptomeningeal disease (LMD) from solid tumors has a dismal prognosis, even following treatment with anti-PD-1 therapy. We performed a phase IB study evaluating the safety of Avelumab with whole brain radiotherapy (WBRT) in LMD (NCT03719768). METHODS:Fifteen patients were enrolled with LMD from breast, lung, nasopharyngeal, ovarian, and pancreatic tumors. Patients were treated with Avelumab with WBRT, with the first infusion of Avelumab starting 14 days pre-WBRT and continuing during and post-WBRT for up to 5 cycles. Primary endpoints were safety and 3-month OS (OS3). Secondary endpoints included assessment of immune cells in the cerebrospinal fluid (CSF) using single-cell RNA-sequencing (scRNA-Seq) pre- and post-last treatment of Avelumab. RESULTS:DLTs occurred in 2 patients, ie, adrenal insufficiency, hypothyroidism, and pneumonitis. Treatment-related toxicities occurred in 5 patients with grade 1/2 and 5 patients with grade 3/4. Immune-related adverse events occurred in 5 patients with grade 1/2 and 3 patients with grade 3/4. The OS3 was 67% (10 of the 15; 95% CI: 38%-84%). Median-OS was 3.85 months (95% CI: 0.9-34.4 months) and median-PFS was 3.85 months (95% CI: 0.9-12.1 months). scRNA-Seq analysis of CSF pre- and post-last-treatment showed Avelumab + WBRT stimulated an adaptive immune response associated with a decrease in regulatory T cells (Tregs), among other changes in the expression of immune checkpoints on CD8 + T cells and macrophages. CONCLUSIONS:The combination of Avelumab and WBRT is safe and demonstrates activity in patients with LMD. The identification of high levels of Tregs and macrophages in the CSF of LMD patients offers future avenues for therapeutic development.

Abstract BACKGROUND Leptomeningeal disease (LMD) is a devastating complication of BC has a dismal prognosis. The CSF of LMD pts have an innate (PMID:34035069), but not adaptive, immune response (bioRxiv 2023.03.17.533041; doi: https://doi.org/10.1101/2023.03.17.533041) that is insufficient to combat LMD. We used IT cDC1 to elicit an adaptive response in LMD and found they were safe, induced a Th1 response, cured most HER2+ LMD, and prevented LMD recurrence (PMCID: 9354231). Responses were CD4+ and B cell dependent. cDC1s induce Th1 responses in HER2/HER3 tumors (PMID: 35710296; PMID: 34785506). And there are trials of cDC1s in BC patients (e.g.NCT03384914, NCT03387553, etc.) We hypothesized that a RP2D would be found and the CSF would be remodeled to have a Th1 immunological profile. METHODS Phase I single-arm, dose escalation multicenter study to establish 1) safety of IT cDC1s, and 2) associations between clinical outcomes & translational CSF studies (e.g. scRNAseq, cytokine arrays). Eligibility includes TNBC or HER+ LMD pts, prior pCSpRT/WBRT, ECOG ≤2, normal organ function, life expectancy of ≥ 8 weeks, and an Ommaya. IT cDC1s were administered weekly x 12 wks at one of 4 dose levels (1 X 106 – 5 X 107 cDC1s) until PD, DLT or withdrawal. Primary Endpoints were 1) safety and DLTs, 2) association between clinical endpoints and immune profiles. We used the BOIN design for the MTD. Response was measured using RANO-LM (PMID: 30715514). DLTs were defined as ≥ gr. 3 not due to LMD, resistant to medical intervention/CSF removal. Final pre & post cDC1 treatment cyto/chemokine & scRNAseq will be determined. As of 04 01 2024, 4 pts were treated [3 pts cohort 1 @ 1 X 106 & 1 pt cohort 2 @ 2 X 106]. NCT05809752. Sponsor: Department of Defense

Melanoma-associated leptomeningeal disease (M-LMD) occurs when circulating tumor cells (CTCs) enter into the cerebral spinal fluid (CSF) and colonize the meninges, the membrane layers that protect the brain and the spinal cord. Once established, the prognosis for M-LMD patients is dismal, with overall survival ranging from weeks to months. This is primarily due to a paucity in our understanding of the disease and, as a consequence, the availability of effective treatment options. Defining the underlying biology of M-LMD will significantly improve the ability to adapt available therapies for M-LMD treatment or design novel inhibitors for this universally fatal disease. A major barrier, however, lies in obtaining sufficient quantities of CTCs from the patient-derived CSF (CSF-CTCs) to conduct preclinical experiments, such as molecular characterization, functional analysis, and in vivo efficacy studies. Culturing CSF-CTCs ex vivo has also proven to be challenging. To address this, a novel protocol for the culture of patient-derived M-LMD CSF-CTCs ex vivo and in vivo is developed. The incorporation of conditioned media produced by human meningeal cells (HMCs) is found to be critical to the procedure. Cytokine array analysis reveals that factors produced by HMCs, such as insulin -like growth factor-binding proteins (IGFBPs) and vascular endothelial growth factor-A (VEGF-A), are important in supporting CSF-CTC survival ex vivo. Here, the usefulness of the isolated patient-derived CSF-CTC lines is demonstrated in determining the efficacy of inhibitors that target the insulin -like growth factor (IGF) and mitogen-activated protein kinase (MAPK) signaling pathways. In addition, the ability to intrathecally inoculate these cells in vivo to establish murine models of M-LMD that can be employed for preclinical testing of approved or novel therapies is shown. These tools can help unravel the underlying biology driving CSF-CTC establishment in the meninges and identify novel therapies to reduce the morbidity and mortality associated with M-LMD.

Abstract BACKGROUND HER2+ leptomeningeal disease (LMD) has a poor prognosis with a high unmet clinical need. A previous phase I/II study revealed intrathecal (IT) trastuzumab to be well tolerated with improved overall survival (OS) compared to historical controls in HER2+ breast LMD. Radiotherapy can improve the flow of IT therapy through the cerebrospinal fluid (CSF) and provide symptomatic relief. The monoclonal antibody pertuzumab is used in conjunction with trastuzumab in the systemic management of HER2+ breast cancer. The phase I portion of the current study used a modified toxicity probability interval-2 (mTPI-2) design to determine 80 mg IT pertuzumab as the recommended phase II dose with a fixed dose of 80 mg IT trastuzumab. METHODS The study is designed as a prospective, single-arm, nonrandomized, open-label, phase II trial of radiation therapy followed by IT trastuzumab/pertuzumab in the management of HER2+ breast LMD. HER2+ LMD patients identified by magnetic resonance imaging (MRI) and/or CSF cytology, ≥18, with a life expectancy > 8 weeks are eligible. Treatment is initiated with radiotherapy, whole brain radiotherapy and/or focal brain/spine radiation followed by IT trastuzumab/pertuzumab. The primary objective of the phase II portion is to evaluate OS following IT trastuzumab/pertuzumab. Secondary objectives involve evaluating the response rate (leptomeningeal and parenchymal), and progression free survival (leptomeningeal and parenchymal). A single-arm two-stage trial is designed using the Restricted-Kwak-and-Jung’s Method. The primary endpoint is one-year OS. An interim analysis will be performed after 20 patients are enrolled with plans to enroll an additional 10 patients if criteria are met. This study is open with 14 patients enrolled at the time of submission. Clinical trial information: NCT04588545.

TPS2090 Background: Leptomeningeal disease (LMD) is a devastating complication of BC where it occurs in approximately 5% of pts and has a dismal prognosis. The CSF of pts with LMD have an innate (PMID:34035069), but not adaptive, immune response (bioRxiv 2023.03.17.533041; doi: https://doi.org/10.1101/2023.03.17.533041) that is insufficient to combat LMD. We used IT cDC1 to elicit an adaptive response in murine LMD and found that IT cDC1s were safe, induced a Th1 response, cured most HER2+ LMD, and prevented LMD recurrence (PMCID: 9354231). This Th1 response was CD4+ > CD8+ T cell dependent and produced marked B cell infiltration in the CSF. Mechanistic studies are in progress. These cDC1s have been shown to induce Th1 responses to HER2 and HER3 tumors and prolong survival in preclinical models (PMID: 35710296; PMID: 34785506). And there are several trials of cDC1s in BC patients that are ongoing (NCT03384914, NCT03387553, NCT05504707, NCT05325632, NCT04348747, NCT05378464). On the basis of our murine studies we hypothesized testing cDC1 therapy in pts that a RP2D would be found and the CSF following cDC1 therapy would be remodeled to have a Th1 adaptive immunological profile. Methods: This is a phase I single-arm, non-randomized dose escalation multicenter study to establish 1) the safety of IT cDC1s in pts with LMD, and 2) associations between clinical outcomes & translational CSF studies such as scRNAseq and cytokine/chemokine arrays. Eligibility includes TNBC or HER+ LMD pts, prior pCSpRT/WBRT, ECOG PS ≤2, cytologic or MRI diagnosis of LMD, normal marrow and organ function, life expectancy of ≥ 8 weeks, and an Ommaya reservoir in place. Exclusions included other treatments to treat LMD ≤ 2 wks or < five ½ lives, > 8mg of DXM/D or equivalent, pregnancy or immunodeficiency syndromes were excluded. IT cDC1s were administered once a week x 12 wks at one of 4 dose levels (1 X 106 – 5 X 107 cDC1s) until PD, DLT or withdrawal during those 12 wks. Primary Endpoints was to 1) determine the safety and DLTs, and 2) association between various clinical endpoints and immune profiles in the CSF. We used the Bayesian Optimal Interval (BOIN) design to find the MTD. Response was measured using RANO-LM (PMID: 30715514). DLTs were defined as ≥ gr. 3 not due to LMD & didn’t respond to medical intervention/CSF removal ≤ 96 hrs. Final pre & post cDC1 treatment cyto-, chemokine profiles and cellular/tumor profiles (scRNAseq) will be determined and associations between these and clinical endpoints will be determined. As of 01 30 2024 the 1st cohort closed with three pts (1 TNBC, 2 HER2+ pts)at the dose of 1 X 106 cDC1s. Clinical trial information: NCT05809752 .

Poor responder LMM CSF vs no LMM CSF significantly altered proteins and their IPA analysis

Leptomeningeal disease (LMD) occurs when tumors seed into the leptomeningeal space and cerebrospinal fluid (CSF), leading to severe neurological deterioration and poor survival outcomes. We utilized comprehensive multi-omics analyses of CSF from patients with lymphoma LMD to demonstrate an immunosuppressive cellular microenvironment and identified dysregulations in proteins and lipids indicating neurodegenerative processes. Strikingly, we found a significant accumulation of toxic branched-chain keto acids (BCKA) in the CSF of patients with LMD. The BCKA accumulation was found to be a pan-cancer occurrence, evident in lymphoma, breast cancer, and melanoma LMD patients. Functionally, BCKA disrupted the viability and function of endogenous T lymphocytes, chimeric antigen receptor (CAR) T cells, neurons, and meningeal cells. Treatment of LMD mice with BCKA-reducing sodium phenylbutyrate significantly improved neurological function, survival outcomes, and efficacy of anti-CD19 CAR T cell therapy. This is the first report of BCKA accumulation in LMD and provides preclinical evidence that targeting these toxic metabolites improves outcomes.

Supplemental Table 2: Sample and Patient Clinical Correlative Information

Supplemental Table 1: Flow Cytometry Antibodies

Supplemental Table 5: Cox proportional hazard model analysis identifies immune cell subpopulations associated with better survival outcomes

TPS1099 Background: HER2+ breast cancer patients with leptomeningeal disease (LMD) represent a poor prognosis population with a high unmet clinical need. Although a multitude of treatment options are available for the management of systemic disease, once metastases travel to the leptomeninges, patients have a lack of treatment options aside from traditional local approaches. Data from a phase I/II study reveals intrathecal (IT) trastuzumab to be well tolerated with improved overall survival (OS) compared to historical controls in HER2+ breast LMD. Radiotherapy can improve the flow of IT therapy through the cerebrospinal fluid (CSF) and provide symptomatic relief. The monoclonal antibody pertuzumab is used in conjunction with trastuzumab in the management of metastatic and localized HER2+ breast cancer. Given the role of radiotherapy in the management of LMD along with the role of pertuzumab in the management of HER2+ breast cancer, there is a strong clinical rationale to combine radiotherapy with IT trastuzumab/pertuzumab in the management of HER2+ breast LMD. Methods: The study is designed as a prospective, single-arm, nonrandomized, open-label, phase I/II trial of radiation therapy followed by IT trastuzumab/pertuzumab in the management of HER2+ breast LMD. HER2+ LMD patients identified by magnetic resonance imaging (MRI) and/or CSF cytology, ≥ 18, with a life expectancy > 8 weeks are eligible. Treatment is initiated with radiotherapy, whole brain radiotherapy and/or focal brain/spine radiation followed by IT trastuzumab/pertuzumab. Safety and feasibility will be monitored by a modified toxicity probability interval-2 (mTPI-2) design. Dose reductions of IT trastuzumab will not be allowed. Once the maximum tolerated dose of IT pertuzumab is determined, the phase II portion of the study will commence to determine OS. Secondary objectives involve defining the CSF pharmacokinetics of IT trastuzumab/pertuzumab, evaluating the response rate (leptomeningeal and parenchymal), and progression free survival (leptomeningeal and parenchymal) following IT trastuzumab/pertuzumab. In the phase 2 portion, a single-arm two-stage trial is designed using the Restricted-Kwak-and-Jung’s Method. The primary endpoint is one-year OS. An interim analysis will be performed after 20 patients are enrolled. This study is open with 1 patient enrolled at the time of submission. Clinical trial information: NCT04588545 .

Abstract BACKGROUND IT trastuzumab is used in the management of HER2+ breast LMD. The monoclonal antibody pertuzumab is used with trastuzumab in the management of metastatic and localized HER2+ breast cancer. METHODS The study is a multi-institutional, phase I/II trial of RT followed by IT trastuzumab/pertuzumab. HER2+ LMD patients with a life expectancy >8 weeks were eligible. Treatment was initiated with RT, whole brain and/or focal brain/spine RT followed by IT trastuzumab/pertuzumab. The dose of IT trastuzumab was fixed at 80 mg with a dose escalation of IT pertuzumab. An mTPI-2 design was used to determine the recommended phase II dose (RP2D). Each cycle was 4 weeks with 2x/week treatment in cycle 1, weekly in cycle 2, and every two weeks thereafter. Clinical trial information: NCT04588545. RESULTS A total of 9 patients were enrolled. Median age was 51 (range: 34-63). Seven patients (78%) were hormone receptor (HR)-/HER2+ and two were HR+/HER2+. Four dose cohorts were treated 10, 20, 40, each with one patient and 80 mg pertuzumab with six patients. Study therapy was well tolerated with no dose limiting toxicities identified. Toxicities at least partially attributable to study therapy included fatigue [grade 1 (n = 2); grade 2 (n = 1)], grade 1 extremity pain (n = 1), and grade 1 headaches (n = 1). Two patients were noted to have infected Ommaya reservoirs after which one patient resumed treatment while one patient stopped therapy after noting stable LMD for approximately 24 months. Median survival has not been reached with a 12 month OS of 86%. Best leptomeningeal response has been a partial response in 4 patients (44%) and stable disease in 5 patients (56%). Six patients continue study therapy currently. CONCLUSIONS Study therapy was well tolerated with a pertuzumab RP2D of 80 mg. Preliminary efficacy appears encouraging and continues to be assessed in the phase II portion.

Kinase Array densitometry quantification

Supplementary Figures 1-11. 1. CSF cytology of leptomeningeal patient cohort. 2. Additional clinical timelines of leptomeningeal patient cohort 3. Patient MRIs 4. RNAseq Workflow 5. Differential responses to BRAFi treatment in CSF 6. RNAseq shows CSF from Patient 4 to elicit transcriptional modulation of melanoma cell response to BRAF inhibition 7. CSF from LMM patients elicits transcriptional changes in melanoma cells 8. CSF protects melanoma cells from BRAF inhibitor therapy 9. CSF activates AKT in WM164 10. No-LMM CSF does not induce expression of TGFbeta or activation of AKT 11. ELISA assay showing the levels of TGFβ1 in CSF specimens from extrordi- nary responder (Patient 1) and non-responders.