Intrathecal pemetrexed for newly diagnosed leptomeningeal metastases: a multicenter, open-label, phase I/II study

Abstract

Purpose

This phase I/II study aimed to determine the maximum-tolerated dose (MTD) of intrathecal pemetrexed (IP) with vitamin supplementation, and to evaluate its safety, feasibility and therapeutic activity for newly diagnosed leptomeningeal metastasis (LM) from solid tumors.

Methods

The phase I study followed the classic 3 + 3 design, with IP dose escalating from 15 mg. The recommended dose was applied in the phase II study. IP was administered first as induction therapy twice weekly for 2 weeks, followed by consolidation therapy, once weekly for 4 weeks, then maintenance therapy once monthly. Vitamin supplementation was initiated prior to the first IP dose. The MTD, adverse events (AEs), overall survival (OS), clinical response rate (CRR) and disease control rate (DCR) were evaluated.

Results

A total of 34 patients were enrolled. In the phase I study, 2 patients in the 20 mg group experienced dose-limiting toxicity, including grade 4 leukopenia and grade 5 chemical arachnoiditis. The MTD was established as 15 mg. Among 28 patients in the MTD group, 24 completed induction therapy, 19 completed consolidation therapy and 11 proceeded to maintenance therapy. The overall AEs rate was 74% (25/34), and severe AEs rate (> grade 3) was 15% (5/34). In the MTD group, the CRR, DCR and median OS were 46% (13/28), 75% (21/28) and 8.1 months (95% CI, 6.5–11.9).

Conclusion

IP at 15 mg, with folic acid and vitamin B12 supplementation initiated before the first IP dose, demonstrated feasibility and exhibited controllable toxicity. This regimen offers a new treatment option with therapeutic activity for newly diagnosed LM patients with solid tumors.

Trial registration

ClinicalTrials.gov Identifier NCT05289908.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11060-025-05160-4.

Introduction

Leptomeningeal metastasis (LM) is a devastating complication of multiple malignancies, characterized by cancer cell involvement in the cerebrospinal fluid (CSF) and leptomeninges, with the typical median overall survival (OS) of only 3–4 months [1]. Approximately 10% of patients with solid tumors develop LM during their disease course, with lung cancer, breast cancer, and melanoma representing the three most common primary malignancies [2]. Intrathecal chemotherapy remains a primary treatment for LM, offering the advantage of direct CSF drug delivery to maximize therapeutic exposure to tumor cells [3]. Compared with systemic administration, low-dose intra-CSF administration can achieve higher CSF drug concentration with a better cytotoxic effect [4]. However, chemotherapeutic agents that can be safely administered intrathecally for the treatment of LM are extremely limited [5]. It is crucial to identify novel agents with both favorable safety profiles and potential activity.

In 2017, we conducted the first study of intrathecal pemetrexed (IP) in the world, which established the feasibility of administering IP in patients with recurrent and refractory LM from lung adenocarcinoma [6]. The recommended dose of IP for LM was 10 mg on a schedule of 1–2 times per week based on pharmacokinetic studies. In that study, prophylactic folic acid and vitamin B12 supplementation was not used, as the intrathecal dose was less than 3% of the systemic dose, potentially impacting treatment efficacy [6]. However, our subsequent research revealed that vitamin supplementation significantly reduced the incidence of hematological toxicity, a primary adverse effect and dose-limiting toxicity (DLT) of pemetrexed administered intrathecally [7]. Consequently, our previous study was limited by an inadequate investigation of patients with recurrent or refractory disease [6]. Also, the maximum tolerated dose (MTD) and safety profile of IP with folic acid and vitamin B12 supplementation initiated before the first IP dose remain undefined, particularly in newly diagnosed LM patients who have not previously undergone intrathecal chemotherapy [6].

Subsequent to our earlier research, several other studies on IP have been conducted, typically employing dosing intervals of once every 1–3 weeks [8–10]. However, the MTD reported in these studies (30–50 mg) was significantly higher than in our previous work, while the incidence of adverse events (AEs) was notably lower [6–11]. The significant variability between studies has caused confusion regarding the treatment regimen and dosage of IP in clinical practice. Therefore, further studies are urgently needed to evaluate its true tolerability, optimal dosage, safety profile, side-effect management, and potential activity of IP using a pharmacokinetics/pharmacodynamics driven regimen based on the previous study on animal models [12].

Based on our previous findings [7], this prospective phase I/II study was designed to determine the MTD, safety, feasibility and clinical activity of IP with folic acid and vitamin B12 supplementation initiated before the first IP dose in patients with newly diagnosed LM from various solid tumors. The dosing regimen in this study maintained the same frequency and intervals for IP as in our previous research. The key difference lies in the dose of IP, which has been adjusted following pre-administration vitamin supplementation and implemented through dose escalation.

Patients and methods

Study design and participants

This was a multicenter, open-label, single-arm phase I/II clinical study (ClinincalTrials.gov identifier: NCT05289908). Eligible patients were aged 18–75 years, had pathologically confirmed malignant solid tumors with normal organ and marrow function, life expectancy of greater than 2 months, and had a newly diagnosed LM confirmed by positive CSF cytology or LM-related specific imaging findings and neurological signs if CSF cytology was negative. The diagnosis of LM was ascertained according to the guidelines by the European Association of Neuro-Oncology–European Society for Medical Oncology (EANO-ESMO) co-published [2]. Hematologic malignancy or primary central nervous system (CNS) malignancy, or other reasons that made them unsuitable for this study, including serious central nervous system disorders, hydrocephalus or other factors suggestive of CSF obstruction, lethal or extensive systemic diseases with few treatment options, psychiatric illness, or poor compliance, were excluded.

Procedures and treatment regimen

Phase I of the study was a 3 + 3 dose escalation trial (Fig. 1). Based on the data of our previous study [6], we chose to treat patients starting at a dose of 15 mg IP with folic acid and vitamin B12 supplementation initiated before the first dose and escalating each dose group by 5 mg (20 mg,…). DLT was assessed during the first 6 weeks of treatment. If 2 or more of the 6 patients experienced a DLT, dosing escalation would cease and MTD would be reached. The MTD, also the recommended dose for phase II, was the next lower dose at which < 2/6 subjects experienced a DLT. The DLT was defined as ≥ grade 3 neurological toxicities (e.g., chemical meningitis) or other ≥ grade 4 toxicity.

Fig. 1.

The flow chart for the trial. LM, leptomeningeal metastasis; IP, intrathecal pemetrexed; CNS, central nervous system; MTD, maximum-tolerated dose; AEs, adverse events

Intrathecal pemetrexed (Alimta, Eli Lilly and Company) was administrated by lumbar puncture, first as induction therapy, twice per week for 2 weeks, followed by consolidation therapy, once per week for 4 weeks. 100 mg of pemetrexed lyophilized powder was dissolved in 50 mL of 0.9% sodium chloride solution to obtain a 2 mg/mL pemetrexed solution. Based on the dose of IP in the study protocol, withdraw the corresponding volume of the prepared solution (e.g., 15 mg is equivalent to 7.5 mL). 1 mL of dexamethasone (5 mg/mL) was administered via intrathecal injection combined with pemetrexed during each IP administration. If LM progresses or DLT occurs during therapy, the treatment will be discontinued. For patients who were evaluated to be effective or stable, maintenance therapy was initiated and administered once per month until the patient’s death, LM progresses, or intolerable severe AEs occurred. The treatment regimen is provided in Supplementary Fig. 1. In addition, folic acid 400 µg was administered orally once daily, prior to the first IP, until 21 days after the last IP. A single dose of vitamin B12 1000 µg was administered by intramuscular injection before the first IP, once every 2 months. Involved-field radiotherapy (IFRT) encompassing whole-brain irradiation, focal intracranial lesions, and/or segmental spinal irradiation was used concurrent with IP in selected patients [3]. Patients who had received systemic treatment prior to enrollment were allowed to continue their original regimens, including tyrosine kinase inhibitors (TKIs) or chemotherapy. However, since these patients developed LM progression while receiving the same therapies, this suggests a failure of the systemic treatments against LM. The observed CNS progression thus reflects the inherent limitations of these systemic regimens in controlling LM. In addition, symptomatic and supportive treatment was permitted for patients with severe conditions.

Endpoints and assessments

As a phase I/II study, it focused on evaluating the feasibility and safety of the treatment protocol. The primary objectives were to determine the MTD of IP with folic acid and vitamin B12 supplementation initiated before the first dose and safety based on the incidence of treatment-related AEs. The secondary objective was OS. In addition, the clinical response rate (CRR) and disease control rate (DCR) were also evaluated. The clinical response was evaluated by RANO-LM working group proposal criteria [13], which were based on three levels of assessment, CSF cytology, neurological status and neuroimaging findings. CRR was defined as the incidence of at least one of the evaluations of CSF cytology, neurological status and neuroimaging findings rated as improved, and there was no disease worsening at the same time. The DCR was defined as the incidence of patients who did not experience disease progression during the initial treatment period (induction and consolidation therapy). Disease progression was defined as the worsening of neuroimaging or neurological dysfunction.

CSF cytology examination was performed at baseline and each time of IP using Thinprep plus Papanicolaou staining during induction and consolidation therapy, and then every month during maintenance therapy. Patients with positive baseline CSF cytology were evaluated as having a cytological response when CSF cytology turned negative for 4 weeks or more after IP. Neurological function examination was performed at baseline, weekly thereafter during induction and consolidation therapies, and then monthly during maintenance therapy to record changes in neurological symptoms and signs. The improvement of neurological function for 2 consecutive weeks was evaluated as effective. Neuroimaging evaluation was performed at baseline, at the end of consolidation therapy, 4 weeks after that, and then every 2–3 months during maintenance therapy. Neuroimaging assessment was performed according to the proposal for a revised Leptomeningeal Metastasis Working Group grid by 3 neuro-radiologists and 2 neuro-oncologists [14].

All patients were followed up until death or for at least 12 months after treatment. AEs were graded according to the Common Terminology Criteria for Adverse Events (CTCAE, version 5.0). A Grade 4–5 was defined as serious AEs. Survival time was defined from the enrollment until death or the last follow-up.

Statistical analysis

Since the CRR varies greatly in the clinical study of LM due to the subjective evaluation of researchers [15], we opted for OS to estimate the sample size for this study. The PASS 15.0 software was used to calculate the sample size and the target number was estimated by the median OS. Based on the previous study [6], we hypothesized H0 as an OS of 3.8 months and H1 as 6.8 months, 28 patients were needed in the Phase II study to achieve 80% power with a predefined one-sided alpha of 5%. The R tools 4.0 software was used for data analyses. Fisher’s exact tests were used to evaluate the difference in CRR between patients with various pathological types. Survival analyses were performed using the Kaplan–Meier method. P < 0.05 demonstrated a significant difference.

Results

Characteristics

Between February 2022 and January 2023, 34 patients (16 males and 18 females) were enrolled sequentially (Fig. 1). Patients’ baseline clinical characteristics and disease-related variables were displayed in Table 1 and Supplementary Table 1. LM diagnosis was confirmed in 97% of patients (N = 33) by positive CSF cytology and probable in 3% (N = 1) by neuroimaging findings and typically LM-related neurological system symptoms and signs according to the EANO-ESMO guidelines [2]. Of all participants, 21% (N = 7) of cases had progressive systemic disease and 76% (N = 26) had stable systemic disease outside the CNS. 76% (N = 26) had received systemic chemotherapy and 15% (N = 5) of patients had received immunotherapy before enrollment. 59% of the participants were NSCLC (N = 20) and 47% (N = 16) received TKIs treatment prior to the diagnosis of LM.

Table 1.

Baseline characteristics of the patients

	All Eligible Patients (n = 34)	Patients in the MTD group (n = 28)
Gender
Male	16 (47%)	12 (43%)
Female	18 (53%)	16 (57%)
Median age (years)	56 (45–68)	56 (45–67)
GCS < 15	7 (21%)	7 (25%)
KPS
< 60	16 (47%)	12 (43%)
≥ 60	18 (53%)	16 (57%)
CSF cytology
Positive	33 (97%)	27(96%)
Negative	1 (3%)	1 (4%)
Neuroimaging features
Positive	31 (91%)	26 (93%)
Negative	3 (9%)	2 (7%)
Onset as LMa	3 (9%)	3 (11%)
Pathological types of primary disease
NSCLCb	20 (59%)	16 (57%)
SCLC	3 (9%)	3 (11%)
Breast cancer	8 (24%)	7 (25%)
Gastric adenocarcinoma	1 (3%)	1 (4%)
Esophagus carcinoma	1(3%)	0
Nasopharyngeal carcinoma	1(3%)	1 (4%)
Elevated intracranial pressure
Yes	16 (47%)	14 (50%)
No	18 (53%)	14 (50%)
Systemic disease
progressive	7 (21%)	5 (18%)
Stable	26 (76%)	22 (79%)
Not observed	1 (3%)	1 (4%)
Breast cancer	8 (24%)	7 (25%)
Triple-negative	2 (6%)	2 (7%)
Her-2 +	4 (12%)	3 (11%)
Luminal A	1 (3%)	1 (4%)
Luminal B	1 (3%)	1 (4%)
NSCLC	20 (59%)	17 (61%)
EGFR mutation	14 (41%)	11 (39%)
ALK mutation	1 (3%)	1 (4%)
ALK&MET mutation	1 (3%)	1 (4%)
Not detected	4 (12%)	4 (14%)
TKI treatment prior to enrollment	16 (47%)	13 (46%)
First-generation EGFR-TKIs	10 (29%)	10 (36%)
Second-generation EGFR-TKIs	3 (9%)	1 (4%)
Third-generation EGFR-TKIs	9 (26%)	8 (29%)
AKL-TKIs	2 (6%)	2 (7%)
MET-TKIs	1 (3%)	1 (4%)
Systemic chemotherapy prior to enrollment	26 (76%)	23 (82%)
Pemetrexed-based chemotherapy	12 (35%)	11 (39%)
Non pemetrexed-based chemotherapy	14 (41%)	12 (43%)
Immunotherapy prior to enrollment	5 (15%)	4 (14%)

MTD, maximum-tolerated dose; GCS, glasgow coma scale; KPS, karnofskyperformance status; CSF, cerebrospinal fluid; LM, leptomeningeal metastasis; NSCLC, non-small cell lung cancer; SCLC, small cell lung cancer; Her-2, human epidermal growth factor receptor 2; EGFR, epidermal growth factor receptor; ALK, anaplastic lymphoma kinase; MET, mesenchymal-epithelial transformation factor; TKI, tyrosine kinase inhibitor

^a LM was the initial manifestation of malignancy

^b Includes lung adenocarcinoma (n = 18), lung squamous cell carcinoma (n = 1) and unable to specify the type of NSCLC (n = 1)

DLT and MTD

In the phase I study, a total of 4 patients were enrolled at the first IP dose level of 15 mg because 1 patient withdrew from the study after the first IP for personal reasons. No DLT occurred. The IP dose was subsequently escalated to the second dose level of 20 mg, and an additional 6 patients were enrolled, 2 of whom experienced DLT. One case of grade 4 leukopenia occurred after the fourth IP (accompanied by grade 2 thrombocytopenia), which was relieved after stopping IP, administering recombinant human granulocyte colony-stimulating factor (rhG-CSF), recombinant human interleukin 11 (rhIL-11), recombinant human thrombopoietin (rhTPO) and increasing the dose of folic acid to 800 µg orally once daily. The other case experienced grade 5 chemical arachnoiditis after the initial IP and died 3 days later. Therefore, the MTD of IP with folic acid and vitamin B12 supplementation initiated before the first IP dose for LM patients was defined as 15 mg, which was also the recommended dose for phase II. Subsequently, an additional 24 patients were enrolled in phase II and received 15 mg of IP. The 4 patients treated at the MTD during phase I were included in the phase II analysis to achieve the pre-specified sample size requirement of a total of 28 cases.

Treatment and feasibility

Of the 34 participants, 28 patients completed induction IP (Table 2). Among the remaining 6 patients who did not complete induction therapy, 2 died 3–5 days following their third IP attributed to LM progression; 1 terminated treatment after developing chemical arachnoiditis after the first IP and 3 withdrew consent for personal reasons. A total of 21 patients completed the consolidation IP, and 12 patients received maintenance IP subsequently. Among 28 patients in the MTD group, 24 completed induction therapy, 19 completed consolidation therapy, and 11 proceeded to maintenance therapy (Table 2).

Table 2.

The treatment of the patients

Treatment	All Eligible Patients (n = 34)	Patients in the MTD group (n = 28)
Induction IP
Completion of induction IP	28 (82%)	24 (86%)
Incompletion of induction IP	6 (18%)	4 (14%)
Consolidation IP
Completion of consolidation IP	21 (62%)	19 (68%)
Incompletion of consolidation IP	13 (9%)	9 (32%)
Maintenance IP	12 (35%)	11 (39%)
Other treatments during maintenance IP
CNS radiation therapy	8 (24%)	6 (21%)
Systemic chemotherapy	3 (9%)	3 (11%)
Pemetrexed + Carboplatin	2	2
Etoposide + Carboplatin	1	1
Targeted therapy	11 (32%)	8 (29%)
Osimertinib	8	6
Afatinib	1	0
Savolitinib	1	1
Furmonertinib	1	1

MTD, maximum-tolerated dose; IP, intrathecal pemetrexed; CNS, central nervous system

Eight patients received concurrent IFRT, including 5 with whole brain radiotherapy and 3 with segmental spinal canal radiotherapy. Fourteen patients maintained their prior systemic therapy during the maintenance therapy, including prior TKIs treatment in 11 cases and prior systemic chemotherapy regimens in 3 cases. The remaining 7 patients with progressive systemic diseases did not receive systemic treatment during IP therapy owing to low KPS score (less than 50), severe neurological status or personal factors.

Safety

All participants were assessed for safety. The overall AEs rate was 74% (25/34) (Supplementary Table 2), and the severe AEs rate (> grade 3) was 15% (5/34, 95% exact CI, 4.95–31.05%). Twenty-one of 28 patients in the MTD group experienced one or more AEs (Table 3). In all participants, the most frequent AEs were hematologic toxicity (53%, 18/34) and elevated hepatic aminotransferases (EHA, 38%, 13/34). Hematologic toxicity occurred mainly after 4–6 times of IP, which was manifested as thrombocytopenia (N = 11) and leukopenia (N = 12). Severe hematologic toxicity occurred in 4 cases including 1 grade 4 leukopenia with 20 mg of IP and 3 grade 4 leukopenia and thrombocytopenia with 15 mg of IP. EHA mainly occurred after 2–5 times of IP. 26% of patients required aminotransferase-lowing drugs for management of EHA. In addition, 3 patients (9%) experienced arachnoiditis, including 1 patient of grade 5 at 20 mg dose level (died within 1 week after abandoning treatment) and 2 patients of grade 2 at 15 mg dose level (completely relieved following glucocorticoid therapy). Radiculitis was noted in 9% (3/34) of patients, and mainly in grade 1 and grade 3. No symptomatic treatment was required and no patient discontinued treatment due to radiculitis. Grade 3 hepatocellular jaundice in 1 patient (3%), grade 1–2 transient cognitive disturbance in 2 patients (6%), grade 3 fatigue in 1 patient (3%), and grade 2 nausea in 1 patient (3%) were also observed in this study.

Table 3.

Adverse events in the MTD group (n = 28)

Adverse events	Grade N (%)
	1	2	3	4	5
Hematologic toxicity	5 (18%)	2 (7%)	6 (21%)	3 (11%)	0
Elevation of hepatic aminotransferases	2 (7%)	3 (11%)	6 (21%)	0	0
Elevation of blood bilirubin	0	0	1 (4%)	0	0
Arachnoiditis	0	2 (7%)	0	0	0
Radiculitis	2 (7%)	0	0	0	0
Transient cognitive disturbance	1 (4%)	1 (4%)	0	0	0
Fatigue	0	0	1 (4%)	0	0
Nausea	0	1 (4%)	0	0	0

MTD, maximum-tolerated dose

Clinical outcomes

According to the response assessment criteria of Response Assessment in Neuro-Oncology (RANO)-LM [13], 15 patients were assessed as a response (11 with improvement of neurological function, 8 with improvement of neuroimaging, and 11 with negative conversion of CSF cytology); 10 patients were assessed as stable; and 4 patients were assessed as LM progressive based on worse neurological function and neuroimaging assessment (Supplementary Table 3). The remaining 5 patients were unevaluable without neuroimaging or CSF cytology re-examination.

Of a total of 29 patients who were available for neurological function assessment, 11 were evaluated as improved, 14 were stable and 4 were worse. Of 27 patients with available CSF cytology data, 11 patients were assessed as having CSF cytological response, and 16 patients were stable. For neuroimaging assessment, 26 patients were available for assessment, including 8 improved, 17 stable/equivocal, and 1 progressive. In addition, 68% (23/34) of patients had improvement of LM-related symptoms.

The overall CRR for all patients was 44% (15/34) and the overall DCR was 74% (25/34) with intention-to-treat (ITT) analysis (Supplementary Table 3). The CRR and DCR for patients in the MTD group were 46% (13/28) and 75% (21/28) respectively. The CRR based on pathological types were 40% (8/20) of NSCLC, 50% (4/8) of breast cancer, and 50% (3/6) of other tumors (including small-cell lung cancer, esophageal cancer, nasopharyngeal cancer, and gastric cancer, Supplementary Table 4). No statistical difference was observed in CRR with various primary tumors (P = 0.899). Notably, the CRR of EGFR-mutated NSCLC patients who developed LM during the course of TKIs treatment was 50% (7/14).

Follow-up and survival

All participants were followed up for at least 12 months until death or March 1, 2024. The median follow-up time was 8.1 months. The median OS for all patients was 8.1 months (95% confidence interval [CI], 6.5–11.3, Fig. 2A), and the 12-month survival rate was 35.3% (12/34). The median OS for patients in the MTD group was 8.1 months (95% CI, 6.5–11.9), and the 12-month survival rate was 35.7% (10/28). The median OS for 15 patients with clinical response was 14.9 months (95% CI, 10.7–17.3), and 3.1 months (95% CI, 2.6–7.1) for 19 patients without clinical response (Fig. 2B, P < 0.001). The median OS of the 12 patients who continued maintenance therapy was 16.9 months (95% CI, 11.6–18.9), and 3.9 months (95% CI, 3.3–7.6) for 22 patients who did not enter the maintenance stage (Fig. 2C, P < 0.001). Median survival for patients with NSCLC, breast cancer, and other tumors was 8.2, 6.3, and 4.8 months respectively, with no statistical difference between the 3 groups (Fig. 2D, P = 0.7). Besides, 14 patients who developed LM during the course of TKIs treatment had a median OS of 11.8 months (95% CI, 7.1–15.7).

Fig. 2.

Kaplan-Meier survival curves showing the survival. (A) Survival for total population; (B) Survival for patients with clinical response and without clinical response; (C) Survival for patients receiving and not receiving maintenance therapy; (D) Survival of patients with different primary tumor types

By the end of follow-up, 85% (29/34) of patients died. Among them, death was attributed to extra-CNS disease progression in 7 patients, to LM progression in 17 patients, to a combination of LM and extra-CNS disease in 2 patients, and not directly related to disease progression in 3 patients (1 treatment-related toxicity and 2 severe intrapulmonary infections).

Discussion

This study demonstrated that the MTD of 15 mg IP with vitamin supplementation, administered on a regimen of twice-weekly dosing for 2 weeks (induction), followed by once-weekly dosing for 4 weeks (consolidation), and then monthly dosing (maintenance), was feasible and exhibited manageable toxicity. Furthermore, this treatment protocol exhibited clinical activity. This study established a feasible IP regimen for newly diagnosed LM from solid tumors.

Pemetrexed, a multitargeted antifolate, has demonstrated broad antitumor activity across diverse solid tumors, including CNS malignancies [16–21]. The anti-tumor efficacy of chemotherapeutic drugs, including pemetrexed, is related to drug concentration, duration of exposure, and rational drug administration regimens, which are mainly influenced by the half-lives of the drugs [22]. A rat model study for IP in 2011 demonstrated sustained high pemetrexed concentrations in CSF following a 1 mg/kg dose, with CSF levels at 24 h post-injection at 0.143 µM, close to the half maximal inhibitory concentration (IC50) observed in human solid tumor cell lines [12]. Based on CSF volume and drug concentration, and considering safety, this study suggests an optimal starting dose of 5–10 mg pemetrexed for human studies. More importantly, half-lives of pemetrexed for the initial distribution/elimination and terminal elimination phases were 0.43 and 1.43 h in rat CSF [12], respectively, which were similar to those of classical intrathecal chemotherapeutic drugs (methotrexate, cytarabine and thiotepa) in the CSF [23].

Several other studies have been conducted, typically employing regimens of IP once every 1–3 weeks [8, 9]. These regimens lack theoretical support from pharmacokinetic data on pemetrexed metabolism in CSF [4, 6, 12]. The human body produces approximately 500 mL of CSF daily, which is twice the total CSF volume in middle-aged adults (250 mL) [24]. These results in a remarkably rapid CSF turnover rate compared to blood renewal. Our prior phase I study of IP showed peak plasma concentrations within 3–12 h, indicated its rapid clearance from the CSF [6]. Given that conventional intrathecal chemotherapeutic agents (excluding sustained-release formulations) exhibit short half-lives in CSF (several hours) [2], these agents including methotrexate, cytarabine, and thiotepa require intensive administration of 12 doses over 8 weeks: twice weekly for 4 weeks (induction phase), followed by once weekly for another 4 weeks (consolidation phase) [2]. This frequency maintains therapeutic drug concentrations for sustained anti-tumor efficacy, as chemotherapy related response depends on both drug levels and exposure duration, especially for cell cycle-specific antitumor drugs including pemetrexed, methotrexate, and cytarabine [4]. Liposomal cytarabine represents the exception, with its sustained-release properties permitting biweekly administration due to an extended 14–21 day CSF half-life [2]. Novel intrathecal therapeutic drugs including topotecan, etoposide, and even trastuzumab and pertuzumab, adhere to conventional intrathecal chemotherapy regimens: induction therapy administered at least twice weekly, followed by consolidation therapy administered at least once weekly. Following induction and consolidation therapy, tumor cells in the CSF were controlled, leading to the initiation of monthly maintenance therapy [25–27]. In our previous phase I study, IP demonstrated a high incidence of hematologic toxicity [6]. Grade 3 or higher hematologic toxicity occurred following intrathecal doses of 10–15 mg administered without vitamin supplementation [6]. Consequently, the pemetrexed induction schedule was shortened to 2 weeks. Otherwise, AEs may delay the initiation of subsequent consolidation therapy. Our subsequent studies, including the current one, indicate that adverse events are controllable when the induction period for IP is limited to 2 weeks. Despite reducing the duration of induction therapy, the total time for induction and consolidation therapy still reached 6 weeks. Throughout the 6 weeks, sustained drug concentrations in the CSF maintained antitumor efficacy, with a clinical response rate of 46% of and a disease control rate of 75% in our present study. Patients then transitioned to monthly maintenance therapy. Additionally, this study revealed that 20 mg of IP could induce grade 4–5 severe adverse events, which is inconsistent with previous research findings and warrants clinical attention. Building upon our previous researches [6, 7], this study further clinically validated a scientifically rational treatment protocol for IP, providing a novel therapeutic option for LM patients with solid tumors.

The incidence of treatment-related toxicity constitutes a primary determinant of drug dosing regimens. Previous studies have found that hematological toxicity and EHA are the main treatment-related AEs of IP [6, 28]. Building upon previous clinical trials [6, 7], this study continued dose escalation of IP with folic acid and vitamin B12 supplementation. The results showed that the supplementation of folic acid and vitamin B12 initiated before the first IP dose elevated the MTD of IP from 10 mg to 15 mg. However, hematological toxicity is still one of the main factors interfering with the completion of treatment. In this study, the incidence of grade 3 or higher hematological toxicity was 32% in the 15 mg dose group. Furthermore, grade 3 EHA was observed in 21% of participants in the 15 mg dose group, which was higher than the previous incidence of 9% at the 10 mg dose level [7], and most of them occurred during the induction IP period (IP, twice a week). Thus, higher dose and frequency of induced IP may increase the incidence of high-grade hematological toxicity and EHA. Compared with conventional systemic administration of pemetrexed, the markedly shortened dosing interval of induction IP may lead to hematopoietic cell depletion and hepatic metabolic impairment. Therefore, in order to avoid the occurrence of severe AEs, the regimen of induction IP is recommended to be twice-weekly administration (maximum 4 doses over 2 weeks). In addition, IP-related chemical arachnoiditis was observed in this study, which has never been reported for systematic chemotherapy with pemetrexed. Obviously, combined with our previous clinical studies [6, 7], this study further confirmed that there was a dose gradient effect in the toxicities for IP, indicating an increase in the incidence of AEs with increasing dosage, particularly the occurrence of severe treatment-related AEs such as neurological toxicity.

In this study, the 15 mg dose regimen yielded a 46% CRR, 75% DCR, and median OS of 8.1 months, which were obviously better than those in our previous study on 10 mg of IP (CRR 31%, DCR 54%, and OS 3.8 months) [6]. These results suggest that with folic acid and vitamin B12 supplementation initiated before the first IP dose, the 15 mg IP might have better activity with controllable AEs. Moreover, this study once again confirmed that IP was effective for LM from various primary solid tumors, similar to previous studies on 10 mg IP combined with IFRT [7], and CRR was not affected by the pathological type of the primary tumor. Moreover, these results surpassed the historical median survival time previously reported for this disease [1].

In recent years, third-generation EGFR TKIs have been approved as the standard first-line therapy for patients with advanced EGFR-mutant NSCLC. Moreover, some studies have demonstrated the efficacy of third-generation EGFR TKIs therapy in LM patients with EGFR-mutant NSCLC [15, 29–33]. However, acquired resistance involving EGFR-dependent or non-EGFR-dependent mechanisms inevitably occurs and hampers their clinical benefits [34–36]. LM exhibits a high incidence in patients with EGFR–mutant NSCLC during the course of third-generation EGFR TKIs treatment. Among 20 LM patients with NSCLC in this study, 14 developed LM during the course of EGFR TKIs treatment, including 9 patients who had received the third-generation TKIs therapy. For those patients, the CRR and median OS were 50% and 11.8 months respectively. These results suggested that IP was effective for LM patients of NSCLC who have previously received TKIs therapy, and we will further validate the activity of IP in NSCLC patients with LM progression after third-generation TKIs treatment (NCT06296745). In addition to studies on NSCLC-related targeted agents for the treatment of LM, some evidence supports the activity of trastuzumab emtansine (T-DM1), trastuzumab deruxtecan (T-DXd), pyrotinib and tucatinib in LM patients with human epidermal growth factor receptor 2 (Her-2) positive breast cancer [37–40]. Whether the combination of targeted agents and intrathecal chemotherapy can further improve the survival of patients deserves exploration in the future.

There were several limitations in this study. First, radiotherapy serves as one of the therapeutic modalities for LM. It is particularly recommended for patients with: imaging-confirmed intracranial parenchymal brain metastases, intraspinal metastatic nodules with significant cauda equina involvement, or impaired cerebrospinal fluid circulation. Radiotherapy may potentially impact AEs and efficacy outcomes. In this study, 8 patients received radiotherapy. Radiotherapy did not increase the incidence of AEs, with no more than grade 3 toxicities observed. It suggested that combining 15 mg IP with radiotherapy maintains a comparable safety profile to the previous study [7]. However, although the survival benefit of radiotherapy for LM remains controversial [41, 42], its impact on treatment efficacy (notably CRR) cannot be disregarded. Further large-scale studies should be required to validate the therapeutic efficacy of IP.

Second, since LM lesions are non-measurable by neuroimaging criteria, traditional response evaluation criteria in solid tumors (RECIST) are inappropriate for assessment [43]. This limitation precludes the use of objective response rate as a primary endpoint and complicates accurate determination of progression-free survival. Although the RANO criteria have been widely adopted, significant variability in CRR reporting persists across LM studies [15]. Consequently, we selected OS as a secondary endpoint with a more objective efficacy measure, while designating CRR as other pre-specified outcome. However, it should be noted that OS interpretation may be confounded by systemic tumor burden and subsequent treatment interventions. Therapeutic evaluation in LM remains particularly challenging due to the disease’s complex nature and multiple confounding factors.

Conclusion

This phase I/II study demonstrated that intrathecal injection of 15 mg pemetrexed with vitamin supplementation was feasible and exhibited a manageable toxicity profile while showing therapeutic activity. It should be noted that the implementation of the treatment regimen requires rigorous monitoring of complete blood count, liver function, CSF biochemical parameters, and neurological signs and symptoms to enable timely detection and management of treatment-related AEs. When combined with rigorous monitoring, these findings provide a potential new treatment option for newly diagnosed LM patients with solid tumors. Moreover, it provides a validated pemetrexed dose and schedule of IP in subsequent trials.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

AEs

Adverse events

CNS

Central nervous system

CRR

Clinical response rate

CSF

Cerebrospinal fluid

DCR

Disease control rate

DLT

Dose-limiting toxicity

IP

Intrathecal pemetrexed

LM

Leptomeningeal metastasis

MTD

Maximum-tolerated dose

OS

Overall survival

Author contributions

Conceptualization: Z. Pan; Project Administration: Z. Pan, G. Yang; Writing-Reviewing and Editing: Z. Pan, G. Yang; Writing-Original Draft Preparation: X. Ye, Y. Huang, P. Tai; Formal Analysis: X. Ye, Y. Huang; Data Curation: Z. Wang; Assisting in Data Curation: M. Liu, L. Shen; Interpretation of Data: X. Pang, T. Yuan; Assisting in interpretation of clinical data: X. Sun, R. Tang, A. Gu. All authors have read and approved the final version of the manuscript.

Funding

This study was supported by grants from Guangdong Basic and Applied Basic Research Foundation (2023A1515140145), Guangzhou Medical University Research Ability Enhancement Project (2024SRP215), and Huizhou Science and Technology Innovation Team Project (2023EQ050012).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics statement

This phase I/II study was conducted according to the guidelines of the Declaration of Helsinki and was approved by the Institutional Review Board of the First Hospital of Jilin University, Changchun, China (Approval number: 2017 − 246). Informed consent was obtained from all subjects. (ClinicalTrials.gov Identifier: NCT05289908).

Conflict of interest

The authors have no conflict of interest to declare.