1047 - Predicting Survival and Treatment Completion of Proton Craniospinal Irradiation in Leptomeningeal Disease: A Prognostic Nomogram
Presenter(s)
T. S. Kanala1, S. Perni2, M. C. Tom2, A. J. Ghia2, D. N. Yeboa2, M. F. F. McAleer2, T. A. Swanson2, S. L. McGovern2, C. Wang2, C. Chung2, B. De2, J. J. Chen2, B. J. O'Brien3, R. K. Murthy4, I. Glitza5, X. Le6, J. Li2, A. S. Gautam7, F. Poenisch7, and T. Beckham2; 1Department of Radiation Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX, 2Department of CNS Radiation Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX, 3Department of Neuro Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX, 4Department of Breast Medical Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX, 5The University of Texas MD Anderson Cancer Center, Houston, TX, 6Department of Thoracic-Head & Neck Medical Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX, 7Department of Radiation Physics - Patient Care, The University of Texas MD Anderson Cancer Center, Houston, TX
Purpose/Objective(s): Phase 2 randomized data demonstrate improved outcomes using proton craniospinal irradiation (pCSI) for patients with leptomeningeal disease (LMD) from breast/NSCLC. However, pCSI is resource-intensive and optimal patient selection is challenging as some patients experience rapid decline and are unable to complete treatment. We hypothesized that baseline clinical factors may identify patients most likely to benefit from pCSI and help develop a prognostic nomogram for patient selection.
Materials/Methods: We retrospectively analyzed a single-institution registry of patients with LMD dispositioned for pCSI (3/2020 - 9/2024). Overall survival (OS) and CNS progression-free survival (CNS-PFS) calculated from LMD diagnosis were estimated using the Kaplan-Meier method. A systemic therapy CNS-activity score ranging between 0-10 was created based on strength of recommendation of systemic therapy regimens for brain metastases and LMD on the National Comprehensive Cancer Network guidelines and was recorded before and within 6 weeks of LMD diagnosis. Univariate (UVA) & multivariable (MVA) Cox proportional hazards models were used to identify prognostic factors for OS. A prognostic nomogram was developed and internally validated using bootstrapping.
Results: Of 97 patients treated, 72 (74%) completed pCSI and 25 (26%) did not. The most common reason for incomplete pCSI was clinical deterioration from LMD. Most (n=73, 75%) had breast or lung primaries, 44 (45%) had prior brain metastases, 54 (56%) had radiologically controlled extracranial disease at presentation, and 71(73%) patients had a Karnofsky performance score (KPS) of =80. Eleven (11%) patients received intrathecal therapy (ITx) with 9(81%) patients starting ITx after pCSI. Median OS was 10 months (95% CI, 6–14), and median CNS-PFS was 8 months (95% CI, 6–11). Median OS in those who completed pCSI was 14 months (95% CI 10–21), and 2 months (95% CI 2–3) in those who did not (p<0.0001); baseline patient & disease characteristics did not differ significantly between these groups. On UVA, age, histology, the presence of targetable driver mutations, use of CNS active systemic therapy, & ITx were found to be significantly associated with improved OS and CNS-PFS. On MVA, increasing age (HR 1.02, 95% CI 1.00–1.04; p = 0.030) and non-NSCLC/breast histology (HR 2.51, 95% CI 1.48–4.27; p = 0.001) were associated with worse outcomes, while greater increase in systemic therapy CNS activity score after pCSI was protective (HR 0.82, 95% CI 0.73–0.92; p = 0.001) for OS. A nomogram incorporating these variables demonstrated moderate discrimination (C-index 0.68) and good internal calibration although survival probabilities were conservatively estimated.
Conclusion: Selected patients with LMD achieved longer survival with pCSI, though treatment attrition was substantial in our cohort. A baseline prognostic nomogram enables individualized OS estimation at presentation and warrants prospective validation to refine patient selection.