Phase II Clinical Trial of Image-Guided Cognitive-Sparing SRS in Patients With Brain Metastases

Abstract

SRS for brain metastases is associated with a high rate (52%-64%) of cognitive deterioration. We performed a phase II clinical trial of critical white matter and hippocampal sparing brain SRS in patients with limited brain metastases, to improve cognitive preservation. We present feasibility of this approach and interim results.Adult patients with limited volume (< 10cc) brain metastases were enrolled on this single institution clinical trial from 2018-2021. Inclusion criteria include: age ≥ 18, expected life expectancy > 6 months, pathologic confirmation of malignancy. Prior WBRT was exclusionary. High resolution volumetric MRI with diffusion tensor imaging was performed pre-SRS and at 1-, 3- and 6-months post SRS. Regions of interest (ROIs) were segmented using robust, automated atlas-based tractography and hippocampal (HC) parcellation methods. 20 white matter (WM) tracts critical for cognitive functioning and fine motor skills, and the HC were designated as ROIs to spare during SRS planning. Knowledge-based SRS planning was employed, and all attempts were made to achieve ROI dose constraints without sacrificing target coverage or other OAR doses. Single fraction max dose constraints were 12 Gy for WM/8.4 Gy for HC (iso-effective doses used for 3 and 5 fraction SRS). A well-validated battery of neurocognitive (NCA) tests, including verbal memory and language, was performed pre and 3 months post SRS. Raw scores were converted to t-scores adjusting for age, sex, and years of education. Decline was considered > 1 standard deviation drop.Median age of cohort (n = 31) was 66 years and 58% were female. Most common primary was NSCLC (39%) or melanoma (23%), and 90% of patients had 1-3 brain metastases. Prescription doses ranged from 21-30 Gy in 1-5 fractions; 10 patients were treated with both single and multi-fraction SRS to different targets. Max dose constraints for ROIs were met on 69% of SRS plans. Across plans, constraints were met for 98% of 595 WM tracts and 97% of 64 HC treated. At 6 months post SRS there were no local failures. To date (median follow up: 6 months), 5 patients have died from systemic disease progression and 1 patient from distant brain disease progression. For patients in whom pre and 3-month NCA scores were available (n = 13), 97% of WM ROIs and 100% of HC ROIs met study constraints. At 3 months, 74% of patients had no decline on HVLT total/delayed recall memory tests. Similarly, 93% and 89% showed no decline on BNT and DKEFS letter fluency language tests, respectively. There was no significant change in NCA test scores in the cohort from pre to 3 months post SRS (paired t-test, P-values > 0.05).Critical WM and HC sparing brain SRS using automated segmentation/planning tools is achievable and feasible, though can depend on proximity to targets. Sparing these critical structures may improve cognitive preservation in domains such as verbal memory and language.S. Unnikrishnan: None. R. Karunamuni: None. G. Kim: None. M. Connor: None. M.A. Salans: None. P. Sanghvi: Speaker's Bureau; Varian, VisionRT.A. Bruggeman: Research Grant; American Society of Clinical Oncology. D.R. Simpson: None. J. Yu: None. A. Reyes: None. A. Stasenko: None. K.L. Moore: Research Grant; Varian Medical Systems Inc. Honoraria; Varian Medical Systems Inc. Travel Expenses; Varian Medical Systems Inc. V. Moiseenko: None. C. McDonald: Research Grant; GE.J.A. Hattangadi-Gluth: Research Grant; NIH/NCI. Honoraria; Varian Medical Systems. Consultant; Varian Medical Systems. Site PI for cooperative group trials at UCSD for CNS; Site-PI; NCCN.