Abstract 1311: Feasibility of plasma ctDNA for detecting brain metastasis-specific alterations in lung cancer.

Abstract Brain metastases (BM) are a frequent and devastating complication of non-small cell lung cancer (NSCLC). Molecular divergence between primary NSCLC and BM is common, highlighting the need for non-invasive approaches to detect BM-specific genomic alterations that could guide precision therapy. To address this, we evaluated the feasibility of circulating tumor DNA (ctDNA) to identify BM-specific or enriched alterations by analyzing matched pre- and post-craniotomy plasma samples and corresponding BM tissue. Twenty-eight metastatic NSCLC BM patients undergoing craniotomy were enrolled. Peripheral blood was collected pre-surgery (Day 0, P1) and post-surgery (≥ 8 days, P2) before systemic therapy initiation, and BM tissue was collected during surgery. Following exclusion of 4 patients with insufficient viable tumor, samples from 24 patients were sequenced using the PredicineCARE NGS assay, and data were analyzed using DeepSea, Predicine's proprietary bioinformatics pipeline. After excluding cases with a tumor fraction 0.005 or failed QC, we analyzed 13 complete sets (BM with matched P1/P2 plasma), 3 partial sets (BM with P1 plasma only) and 7 BM tissue-only samples (23 BM samples total). The most frequently detected BM alterations were TP53 (22/23, 96%), PIK3CA (12/23, 52%), RB1 deletion (del)/mutation (mut), EGFR amplification (amp) and CDKN2A/B del (each in 11/23, 48%). Additional alterations were found in BRCA1/2 (10/23, 44%), MYC (9/23, 39%), NF1 (8/23, 35%), and PTEN or BAP1 (7/23, 30%). MET amp, AKT3 amp, or KRAS mut were each found in 6/23 patients (26%). Targetable alterations, including EGFR (ex19del, L858R, ex20ins), and KRAS (G12C) were each detected in 3 patients (13%). While many alterations were shared between BM tissue and the two plasma samples, complete loss of an alteration in post-craniotomy P2 plasma that was detected in both P1 and BM was rare, occurring in only 1/13 patients (EGFR amp). However, a decrease of 50% variant allele frequency was seen for multiple variants in an additional 2/13 patients despite similar tumor fractions, suggesting that detecting ctDNA shed from BM is possible. Notably, NF1, KEAP1 and CD274 del and ERBB2 amp was only detected in BM and only 1/6 MET amps were detected in plasma. Interestingly, one patient who received post-craniotomy brain radiation therapy (RT) one day before P2 collection exhibited new MET and MYC amps and TP53 del in P2 plasma. These variants were initially detected in BM but absent in P1, suggesting that RT may enhance tumor DNA shedding. These findings indicate that reliable ctDNA detection of BM-specific alterations may require assays with greater sensitivity, as well as optimized sample collection timing, such as post-RT, to improve yield and clinical utility. To our knowledge, this is the only study that includes pre-/post-craniotomy samples along with matched BM tissue to directly assess the ability of ctDNA to detect BM-specific alterations. Citation Format: Laura P. Stabile, Khvaramze Shaverdashvili, Purva H. Rumde, Xiaxuan Wu, Autumn Gaither Davis, Pascal O. Zinn, Ahmed Habib, Neslihan N. Gecici, Rener Zhang, Pan Du, Shidong Jia, Yong Huang, Gabriel Sica, Timothy F. Burns. Feasibility of plasma ctDNA for detecting brain metastasis-specific alterations in lung cancer abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 1311.