Leptomeningeal disease, in which malignant cells spread through the cerebrospinal fluid and the leptomeningeal space, affects roughly 5 to 15 percent of patients with advanced solid tumors and carries a median survival typically measured in months. Breast cancer, lung cancer and melanoma are the most common solid tumor origins, accounting for an estimated 12 to 35 percent, 9 to 25 percent and 5 to 25 percent of cases respectively, and the projected annual incidence in the United States is around 110,000 cases. The blood-brain barrier and the blood-cerebrospinal fluid barrier limit the penetration of systemically administered drugs into the intrathecal space, which is the theoretical rationale for delivering chemotherapy directly into the cerebrospinal fluid.

The research team used the TriNetX Research Network, a federated platform of de-identified electronic health records from healthcare organizations worldwide, querying records from August 12, 2005 through August 12, 2025. They identified 2,603 adults with a coded diagnosis of leptomeningeal disease, of whom 292 had received intrathecal chemotherapy plus systemic chemotherapy and 2,311 had received systemic chemotherapy alone. To reduce confounding, the investigators excluded any patient who received radiotherapy to any body site after the leptomeningeal disease diagnosis, then applied propensity score matching on age, sex, race, primary cancer type, prior surgery, concurrent glucocorticoid use and baseline laboratory values, yielding 247 patients in each arm.

After matching, the cohorts were closely balanced: mean age differed by about one year, hematologic malignancies accounted for roughly 71 percent in both groups, breast cancer for 20 to 24 percent, lung cancer for about 8 percent and melanoma for 6 to 7 percent, with absolute standardized differences below 0.11 for most covariates. Mean follow-up was 225.7 days in the intrathecal chemotherapy group and 201.4 days in the control group. The database defined leptomeningeal disease by diagnosis code rather than by a specific diagnostic modality, so the proportions confirmed by cerebrospinal fluid cytology versus imaging could not be determined.

The central finding was a time-dependent pattern. In the matched unadjusted analysis, 30-day mortality was lower with intrathecal chemotherapy, 6.5 percent versus 9.7 percent, although the hazard ratio of 0.64 was not statistically significant. At 90 days the curves were essentially identical, and at 180 days the difference again was not significant. By one year, however, the direction had reversed: 111 deaths, or 44.9 percent, had occurred in the intrathecal chemotherapy group compared with 84 deaths, or 34.0 percent, among controls, a hazard ratio of 1.34 that was statistically significant. Overall mortality across the full follow-up did not differ significantly, with a hazard ratio of 1.26.

In the larger unmatched cohorts, the authors fitted multivariable Cox models adjusting for the same covariates. There, intrathecal chemotherapy plus systemic chemotherapy was associated with a 55 percent reduction in the hazard of death at 30 days, a hazard ratio of 0.449, with a similar but non-significant trend at 90 days. That apparent protection vanished at 180 days, at one year and over the entire follow-up, where the hazard ratios hovered near or slightly above 1.0. The authors argue that the replication of this early-benefit-then-reversal trajectory across both adjusted and matched analyses strengthens the internal validity of the observation, while emphasizing that it reflects association in retrospective data, not proof of causation.

The safety signal was more consistent than the survival signal. Chemotherapy-related adverse effects, such as cytopenia, mucositis, hepatotoxicity and infection, were more than twice as common in the intrathecal chemotherapy group: 23.9 percent versus 10.1 percent at 30 days, 44.9 percent versus 21.1 percent at one year, and 46.2 percent versus 22.3 percent overall, with hazard ratios around 2.5. All-cause meningitis occurred in 36.8 percent of the intrathecal group versus 21.5 percent of controls, and hydrocephalus was roughly doubled in the early post-treatment period, affecting 8.5 percent versus 4.0 percent. Cognitive-communication deficits were also more frequent, 26.3 percent versus 16.2 percent, which the authors suggest may reflect direct neurotoxicity beyond the baseline risk imposed by the disease itself. Sepsis rates were numerically higher but not statistically significant, and seizure rates did not differ.

Subgroup analyses sharpened the picture. When the intrathecal agents were examined separately, neither intrathecal methotrexate nor intrathecal cytarabine showed a significant effect on mortality at any time point, yet both were consistently associated with higher rates of chemotherapy-related adverse effects, and intrathecal cytarabine carried a persistently increased risk of meningitis from 90 days onward. In a subanalysis restricted to patients with lymphoid or hematopoietic malignancies, 129 patients per arm after matching, the elevated one-year mortality signal disappeared entirely, but the excess chemotherapy-related adverse effects persisted, with hazard ratios between 2.89 and 3.44. The authors note that the longer-term mortality disadvantage in the full cohort therefore appears to be driven largely by patients with solid tumors, and they could not perform a corresponding solid-tumor-only subanalysis because too few eligible patients remained.

These database findings align with the sparse prospective literature. The authors reviewed 13 phase 2 or 3 trials of intrathecal chemotherapy in adults with leptomeningeal disease from solid tumors, most frequently breast and lung carcinoma. In those trials, response rates ranged from 0 to 94 percent, median overall survival from 1.8 to 8.2 months, and adverse events such as leukoencephalopathy, headache, confusion and infection were common. Only two of the reviewed trials compared intrathecal chemotherapy against regimens without it. A randomized study in breast cancer by Boogerd and colleagues found median survival of 18.3 versus 30.3 weeks favoring the non-intrathecal arm, with significantly more neurological complications, and a phase 3 trial of intrathecal liposomal cytarabine by Le Rhun and colleagues found no significant overall survival difference. Neither older landmark randomized trial of intrathecal agents included a systemic-chemotherapy-alone control arm, a gap the authors describe as persisting for more than two decades.

The authors conclude that routine use of cytotoxic intrathecal chemotherapy across unselected patients with leptomeningeal disease is unlikely to confer a durable survival benefit and may increase clinically meaningful complications, supporting a more selective, phenotype-driven approach that weighs cerebrospinal fluid flow dynamics, disease pattern, performance status, extracranial disease control and goals of care. They acknowledge important limitations: the diagnosis was code-based, some complications may predate treatment, glucocorticoid exposure was captured only as a binary variable, adjustment for clustering by healthcare organization was not feasible, and excluding radiotherapy may limit generalizability because radiation is frequently part of standard care, particularly in solid tumors. They call for pragmatic comparative studies within biologically defined subgroups, integration of central-nervous-system-penetrant systemic therapies, and patient-centered endpoints such as neurological function and quality of life alongside survival.

Subject of Research: Comparative effectiveness and safety of intrathecal chemotherapy plus systemic chemotherapy versus systemic chemotherapy alone in leptomeningeal disease

Article Title: Efficacy and safety of intrathecal chemotherapy in leptomeningeal disease: a global retrospective multicenter analysis and review of the literature

Article References: Msherghi, A., Patel, C. B., Gule-Monroe, M., Khalaf, A., Eldaya, R. W., Kamel, P. I., Ucisik, F. E., Najah, Q., Ballester, L. Y., Elhadi, M., Wintermark, M., & Alsharedi, M. (2026). Efficacy and safety of intrathecal chemotherapy in leptomeningeal disease: a global retrospective multicenter analysis and review of the literature. eClinicalMedicine, 101, Article 104274. https://doi.org/10.1016/j.eclinm.2026.104274

Image Credits: AI Generated

DOI: 10.1016/j.eclinm.2026.104274

Keywords: leptomeningeal disease, intrathecal chemotherapy, systemic chemotherapy, TriNetX, propensity score matching, overall survival, meningitis, hydrocephalus, neurotoxicity, breast cancer, lung cancer, hematologic malignancies