Abstract A058: Integrating DNA-RNA sequencing analysis to identify low-variant allele frequency fusions in glioblastoma

Abstract The goal of this study was to test whether integrating somatic structural variant (SV) calls from whole-genome sequencing (WGS) with RNA-sequencing (RNA-seq) fusion calls and differential expression (DE) can recover clinically relevant low variant-allele-frequency (VAF) rearrangements in glioblastoma (GBM).We analyzed 10 GBM samples with paired tumor-normal short read WGS and tumor bulk RNA-seq, providing orthogonal profiling of somatic SVs and transcription in GBM tissue. WGS SVs were called with SvABA and annotated using GENCODE. RNA-seq reads were aligned via STAR; fusions were called with Arriba and STAR-Fusion. For each fusion, we performed gene-aware, windowed breakpoint matching to identify DNA breakends within ±100 kb of the fusion breakpoint on the same chromosome pair and stratified support into confidence tiers based on coordinate concordance and gene proximity. We defined RNA-consensus fusions as Arriba∩STAR-Fusion calls and DNA-supported fusions as RNA fusions with an orthogonal SvABA breakend match within ±100 kb. Transcriptomic impact was assessed by DESeq2 (incorporating publicly-available transcriptome profiles garnered from GTEx brain controls), testing enrichment of DNA-supported fusion partners among DE genes (FDR<0.05), and Hallmark gene set enrichment analysis (GSEA). VAF for fusion-linked SVs was estimated through SVTyper.Across the cohort, we observed 1,561 unique strand-aware fusion events, of which 138 (8.8%) had DNA support within ±100 kb (49; 3.1% at ±10 kb) and 30/138 (22%) DNA-supported events called by both Arriba and STAR-Fusion. RNA-consensus fusions were enriched for orthogonal DNA support compared with single-caller fusions (63.6% vs 7.3%, RR=8.7) while DNA-supported events were more likely to show a transcriptional consequence than fusions without DNA support (≥1 DE gene 93.9% vs 66.7%; Fisher’s exact test, OR 7.67, p=0.0014). Overall, 101/138 (74.2%) of DNA-supported fusions had ≥1 DE gene (FDR<0.05). Hallmark GSEA identified 21 significantly enriched pathways (FDR<0.05), dominated by cell-cycle regulators (E2F targets; G2M checkpoint), with additional interferon-γ and epithelial–mesenchymal transition signals, of which 6 (28.6%) included at least one low-VAF fusion-associated gene (VAF<0.2). SV VAF was predominantly low among concordant DNA-RNA fusion-linked events: 73.3% of unique events (n=30) had VAF<0.2, suggesting that many of these rearrangements occur at frequencies consistent with heterogeneous subclonal populations.We highlight concordant DNA-RNA fusion-linked SVs with expression impact, including genes such as CDC25C, COL4A2, and EGFR as candidates for follow-up. Our integrated DNA-RNA framework identifies genomically supported fusion events with transcriptomic impact, enabling the interpretation of biologically relevant low-VAF rearrangements that may reflect subclonality while highlighting candidates of GBM tumorigenesis. Together, these results suggest that DNA-RNA integration can improve sensitivity and support the recovery of low-VAF events compared with single-modality approaches. Citation Format: Shray Parimoo, Eeshaan Rehani, Suraj Rajendran, David C. Wilkes, Michael Sigouros, Eda Nur Kozan, Juan Miguel Mosquera, Andrea Sboner, Olivier Elemento, Iman Hajirasouliha. Integrating DNA-RNA sequencing analysis to identify low-variant allele frequency fusions in glioblastoma [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Brain Cancer; 2026 Mar 23-25; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86(6_Suppl):Abstract nr A058.