2003 - Predictors of Outcomes Following Proton Craniospinal Irradiation for Solid Tumor Leptomeningeal Metastasis

Purpose/Objective(s): Proton craniospinal irradiation (pCSI) improved overall survival (OS) compared to traditional involved-field radiotherapy for patients with solid tumor leptomeningeal metastasis (LM) in a Phase II trial with the greatest advantage observed in lung cancer patients. However, clinical and disease-specific factors predicting which patients derive the greatest therapeutic benefit remain poorly defined.

Materials/Methods: We retrospectively reviewed 54 patients with solid tumor LM treated with pCSI at our institution (2022–2026). All patients received 30 Gy (RBE) in 10 fractions with intensity-modulated proton therapy (IMPT). OS and central nervous system progression-free survival (CNS-PFS) were estimated from the completion of pCSI using the Kaplan-Meier method and compared via log-rank testing. Univariable/multivariable Cox proportional hazards models, Chi-square/Mann-Whitney U and logistic regression analysis were utilized to evaluate the prognostic impact of age, Karnofsky Performance Status (KPS), primary histology, prior CNS radiotherapy, CNS-penetrant agent after pCSI, interval from initial/CNS/LM diagnosis to pCSI, and control of extracranial disease.

Results: Median patient age was 62 years (interquartile range [IQR]: 53–69), and the cohort was predominantly female (80%) with a median baseline KPS of 80 (IQR: 80–90). Breast cancer (59%) and non-small cell lung cancer (NSCLC) (26%) were the most common primary malignancies. Notably, 56% of the cohort had undergone previous CNS radiotherapy. The median interval from LM diagnosis to the pCSI completion was 20 days (IQR: 14–44). With median follow-up (FU) of 6.4 months, median OS and CNS-PFS were significantly longer in patients with NSCLC (22.1 and 14.0 months, respectively) compared to those with breast cancer (7.6 and 6.0 months) and other primary histologies (2.9 and 2.9 months) (p=0.008 for OS; p=0.004 for CNS-PFS). Age, primary histology, and CNS-penetrant agent use emerged as independent predictors of CNS-PFS on multivariable analysis, p=0.038, p=0.01, and p=0.005, respectively. To identify predictors of better survival (>90 days), we analyzed patients with known OS or FU data >3 months (n=52), and only primary histology (NSCLC) remained significant in logistic regression analysis, p=0.046 (HR: 0.167 95% CI: 0.029-0.967).

Conclusion: In this analysis, pCSI conferred longer CNS-PFS and OS in patients with NSCLC vs. breast and other malignancies. Moreover, use of CNS-penetrant agents and age were significantly associated with CNS-PFS. Incorporating biologically-stratified evaluations in future clinical trials will be crucial to better define the role of pCSI and optimize patient selection.