280 - Connectome-Informed Cognition-Sparing Stereotactic Brain Radiosurgery: Comparison of Cognitive Outcomes to Historical Controls

02:15pm - 02:25pm ET

Room 162

Presenter(s)

Andrew Puett, MD, PhD - UC San Diego, La Jolla, CA

A. C. Puett¹, J. S. Kohli², R. Karunamuni³, A. B. Hopper¹, G. G. Y. Kim¹, J. Villalobos¹, N. Menendez¹, J. Helm⁴, K. R. Tringale¹, V. Moiseenko¹, P. Sanghvi¹, C. McDonald^1,2, and J. A. Hattangadi-Gluth¹; ¹Department of Radiation Medicine and Applied Sciences, University of California San Diego, La Jolla, CA, ²Department of Psychiatry, University of California San Diego, La Jolla, CA, ³Research Service, VA San Diego Healthcare System, San Diego, CA, ⁴Department of Psychology, San Diego State University, San Diego, CA

Purpose/Objective(s):

To compare 3-month cognitive outcomes in patients with brain metastases treated by connectome-informed, cognition-sparing stereotactic radiosurgery (COG-SRS) to an historical cohort that received SRS with standard planning.

Materials/Methods:

Patients with brain metastases completing at least one paired baseline and 3-month neuropsychological test (NPT) shared across the phase II COG-SRS trial and Alliance phase III trials of standard SRS (N0574 and N107C) were included for analysis. N0574 and N107C were combined into a pooled Alliance cohort. All NPT scores were grouped into pre-specified cognitive domains of memory, language, processing speed, executive function, and fine motor control. Missing NPT values were handled using multiple imputation (20 datasets). Within each imputed dataset, NPT raw scores were transformed to z-scores using the pooled baseline mean and SD for each test across all trials, with directionality aligned such that higher z-scores indicated better performance. Domain scores were calculated as the mean of aligned z-scores within each domain. An all-domains composite was estimated with equal domain weighting. For each domain, linear mixed-effects (LME) ANCOVA modeled 3-month performance with trial group (COG-SRS vs. Alliance) as the primary predictor, adjusting for baseline cognition and clinical covariates (age, sex, number of metastases, ethnicity, and race), with a subject-level random intercept. Effects were pooled across imputations using Rubin’s rules.

Results:

A total of 171 patients had paired baseline and 3-month neurocognitive testing (Alliance n=114; COG-SRS n=57). In covariate-adjusted mixed-effects modeling, COG-SRS was associated with significantly better 3-month all-domains composite performance compared with Alliance (ß=0.277, 95% CI 0.065-0.489; p=0.01). In domain-specific analyses, the estimated effect favored COG-SRS for processing speed, executive function, language, and fine motor control, reaching statistical significance for language (ß=0.254, 95% CI 0.025-0.483; p=0.03). No difference was observed for memory (ß=-0.067, 95% CI -0.279-0.144; p=0.532).

Conclusion:

Patients with brain metastases who received COG-SRS had significantly better overall cognitive performance at 3 months compared with an historical cohort treated with standard SRS. Domain-level outcomes generally favored COG-SRS, with the exception of memory. These findings establish the benefit of sparing cognition-critical brain networks and point to memory as a uniquely vulnerable domain in this population. Connectome-sparing optimization can translate into measurable early cognitive preservation and is being planned for study in a randomized trial. (NCT05801133)