Highlight
• In a randomized phase 2 trial (NCT04343573), proton craniospinal irradiation (pCSI) prolonged median central nervous system progression‑free survival (CNS‑PFS) to 8.2 months versus 2.3 months with involved‑field photon radiotherapy (IFRT) (P < .001).
• Median overall survival (OS) was significantly longer with pCSI (11.3 months) than IFRT (4.9 months) (P = .04).
• The trial was stopped early for efficacy at interim analysis; results support considering pCSI for eligible patients with leptomeningeal metastasis when available.
Background: clinical context and unmet need
Leptomeningeal metastasis (LM) occurs when systemic cancer cells seed the cerebrospinal fluid (CSF) and leptomeninges. LM carries a poor prognosis, typically measured in weeks to a few months, and causes progressive neurologic decline. Therapeutic options are limited and heterogeneous: intrathecal chemotherapy, systemic therapy (including targeted agents that penetrate CSF), and focal radiotherapy to symptomatic sites. Whole neuraxis irradiation may be clinically appropriate for extensive LM but is technically challenging with photon techniques because of high integral dose and toxicity.
Involved‑field photon radiotherapy (IFRT) — treating symptomatic bulky disease or focal sites — remains a common palliative approach. Proton craniospinal irradiation (pCSI) offers theoretical advantages for LM because of the physical properties of protons (Bragg peak) that allow better sparing of off‑target tissues, potentially enabling safer whole‑neuraxis dosing and reduced marrow exposure. However, prospective randomized data comparing pCSI with IFRT have been lacking.
Study design
This randomized, phase 2 trial (ClinicalTrials.gov Identifier: NCT04343573) was conducted from April 16, 2020, through October 11, 2021. The trial enrolled patients with LM arising from non–small cell lung cancer (NSCLC) and breast cancer; patients with other solid tumors were recruited into an exploratory pCSI cohort. Participants were stratified by histology and systemic disease status and randomized 2:1 to receive pCSI or IFRT.
The primary endpoint was central nervous system progression‑free survival (CNS‑PFS). Secondary endpoints included overall survival (OS). The randomization yielded 63 randomized patients (42 assigned to pCSI and 21 to IFRT) and an additional exploratory pCSI cohort of 35 patients (total N = 98). Baseline characteristics included a predominance of female participants (72/98, 73.5%) and median age of 59 years (IQR 50–65).
Key findings
Primary outcome — CNS‑PFS: At final analysis, pCSI produced a substantial and statistically significant benefit in CNS‑PFS compared with IFRT. Median CNS‑PFS was 8.2 months (95% CI, 6.6–15.3) in the pCSI arm versus 2.3 months (95% CI, 1.2–4.0) in the IFRT arm (P < .001). The magnitude of this difference (median improvement of approximately 6 months) is clinically meaningful in the context of LM, where natural history without effective CNS control is short.
Secondary outcome — overall survival: The trial also reported a statistically significant OS advantage for pCSI. Median OS was 11.3 months (95% CI, 7.5–18.3) with pCSI versus 4.9 months (95% CI, 3.9–15.0) with IFRT (P = .04). This OS benefit suggests CNS disease control translated into a survival advantage rather than merely a delay in neurologic progression.
Exploratory cohort: Among the 35 patients treated with pCSI in the exploratory cohort (other solid tumor primaries), median CNS‑PFS was 5.8 months (95% CI, 4.4–9.1) and median OS was 7.0 months (95% CI, 5.4–10.6). These results indicate the benefit of pCSI may extend beyond breast and NSCLC primaries, although the trial was not powered to make definitive histology‑specific comparisons in that cohort.
Trial conduct: The trial was stopped early after a planned interim analysis showed a significant benefit for pCSI, meeting stopping criteria for efficacy. Early stopping strengthens ethical justification but can overestimate treatment effect sizes; the final analysis continued to show consistent benefit.
Safety and toxicity: The summary provided does not detail adverse events, treatment‑related morbidity, or quality‑of‑life outcomes. Craniospinal irradiation (CSI) — whether photon or proton — can be associated with hematologic suppression, nausea, fatigue, alopecia, and potential neurocognitive effects from cranial components. Proton delivery typically reduces integral dose to bone marrow and visceral organs and may lower certain toxicities (for example, less gastrointestinal or marrow exposure), but the specific safety profile in this randomized cohort should be reviewed in the full manuscript for precise rates of grade ≥3 events, rates of treatment interruption, and late effects.
Interpretation and clinical significance
These results are practice‑informing. A median CNS‑PFS of 8.2 months and OS of 11.3 months in a randomized population with LM reflect a meaningful shift compared with historical LM outcomes and the control arm. For patients with LM from breast cancer or NSCLC who are eligible for craniospinal irradiation, pCSI appears to offer superior CNS control and survival compared with focal IFRT.
Key considerations for clinicians:
• Patient selection: pCSI is most relevant for patients with diffuse or extensive LM involvement where a neuraxis approach is rational. Patients’ performance status, systemic disease control, life expectancy, and concurrent systemic therapy options (including targeted agents with CSF activity) must inform individualized decisions.
• Multidisciplinary care: Integration with medical oncology (systemic therapies), neuro‑oncology, radiology, and supportive care is essential. pCSI may be combined sequentially or concurrently with systemic agents in selected cases, but drug–radiation interactions and cumulative toxicities require oversight.
• Access and logistics: Proton therapy availability is limited in many regions and cost is higher relative to photon therapy. Center experience with CSI planning and delivery, immobilization, and management of hematologic effects is necessary to reproduce trial outcomes.
Expert commentary and limitations
Strengths of the trial include randomized design, stratification by key prognostic factors (histology, systemic disease status), and an actionable primary endpoint that reflects CNS disease control. Demonstration of an OS benefit strengthens the case that improved CNS control has meaningful downstream effects.
Limitations and caveats:
• Phase 2 and sample size: Although randomized, the phase 2 design and modest sample size warrant cautious interpretation. Early stopping for efficacy can result in effect size inflation; confirmatory larger phase 3 data would be ideal.
• Generalizability: The study was conducted at centers with proton capability and specialized expertise. Results may not generalize to settings without experienced proton teams or to patients with very poor performance status who were likely underrepresented.
• Detailed toxicity, quality of life, and neurocognitive outcomes were not reported in the provided summary. These are crucial for weighing benefits against risks, especially for patients with limited life expectancy.
• Interaction with systemic therapies: The summary does not report contemporaneous use of targeted therapies (e.g., EGFR or HER2 agents) or immune checkpoint inhibitors; such interactions could modify outcomes. Subgroup analyses based on systemic therapy use and molecular subtype would be informative.
Mechanistic plausibility: pCSI’s physical dosimetry enables irradiation of the entire neuraxis while reducing dose to non‑target tissues (e.g., anterior organs and marrow). Improved CNS disease eradication plausibly reduces neurologic progression, preserves performance status, and allows continued systemic therapy, which together can extend survival.
Practical implications and next steps
For clinicians managing LM, this randomized evidence supports offering pCSI when whole‑neuraxis irradiation is considered and proton therapy is available. Decision‑making should incorporate disease distribution, expected benefit margin, patient goals, performance status, and resource implications.
Important next steps include:
• Publication of full safety and quality‑of‑life data to allow balanced risk–benefit assessment.
• Larger, multicenter confirmatory trials or pooled analyses to validate effect size and explore histology‑specific and systemic therapy–interaction effects.
• Cost‑effectiveness analyses and health‑systems planning to expand equitable access to proton therapy where indicated.
Conclusion
This randomized phase 2 trial demonstrates that proton craniospinal irradiation significantly improves CNS‑PFS and OS compared with involved‑field photon radiotherapy in patients with LM from solid tumors (notably breast cancer and NSCLC). The findings support considering pCSI for appropriate patients when available, with multidisciplinary assessment and close attention to safety and quality‑of‑life outcomes. Confirmation in larger studies and transparent toxicity reporting will guide broader adoption and policy decisions.
Funding and trial registration
Trial registration: ClinicalTrials.gov Identifier: NCT04343573.
Primary trial report: Yang JT, Yerramilli D, Pentsova E, Wolden S, Young RJ, Correa DD, Imber BS, Wijetunga NA, Goglia AG, Zhang Z, Zheng J, Baser R, Bernstein A, Kratochvil L, Xiao J, Hattangadi‑Gluth J, Miller AM, Wilcox JA, Betof Warner A, Yu H, Kris MG, Seidman AD, Powell SN, Boire A. Proton Craniospinal Irradiation for Patients With Leptomeningeal Metastasis: A Randomized Clinical Trial. JAMA Oncol. 2025 Nov 1;11(11):1293-1301. doi: 10.1001/jamaoncol.2025.3007. Erratum in: JAMA Oncol. 2025 Nov 1;11(11):1398. doi: 10.1001/jamaoncol.2025.4502. PMID: 40906462; PMCID: PMC12412039.
References
Yang JT, Yerramilli D, Pentsova E, et al. Proton Craniospinal Irradiation for Patients With Leptomeningeal Metastasis: A Randomized Clinical Trial. JAMA Oncol. 2025;11(11):1293-1301. doi:10.1001/jamaoncol.2025.3007. ClinicalTrials.gov Identifier: NCT04343573.