Article Text
Abstract
Introduction Brain arteriovenous malformations (bAVMs) are dysplastic tangles of blood vessels with direct arterial-venous shunting. Evidence has suggested that bAVMs may arise from different somatic or germ-line mutations. However, there are no methodologies to molecularly characterize bAVMs without open surgery. Whether different molecular subgroups may explain divergent natural histories remains unknown. We describe a novel endovascular biopsy (EB) technique to permit comprehensive genome-wide transcriptomic classification of bAVMs without need for open surgery in living patients.
Methods EB and computational fluid dynamic modeling (CFD) was performed during preoperative embolization. Cells were obtained from a bAVM juxta-nidal feeding artery and systemic iliac artery (control) with a platinum detachable coil and 0.035-inch wire, respectively. Endothelial cells (ECs) were isolated with fluorescence assisted cell sorting (FACS) and sequenced on an Illumina HiSeq 4000 sequencer. Differential gene expression and gene ontology analyses were performed.
Results EB was successful in 4 patients without complication. FACS sorting detected viable ECs (CD31-positive, DAPI-negative) in each patient. EB yielded 269.0 ± 79.9 cells per biopsy (control: 309.2 ± 86.6 cells; bAVM: 228.8 ± 133.4). Differential gene expression analysis demonstrated 106 differentially expressed genes in bAVM ECs (FDR < 0.05) (figure 1A), and clear separation of bAVMs and patient-matched controls on principal component and heatmap analyses (figure 1B). Differentially expressed genes were enriched for bAVM-related Ras-MAPK signaling and angiogenic cell motility pathways (p<0.05) (figure 1C). When compared to open surgery, EB detected 83.3% of genes and genome-wide expression strongly correlated with ECs from open surgery (R2=0.8). CFD modeling of blood flow, including wall shear stress and oscillatory shear index, correlated with EB gene expression and supported evidence of vascular remodeling.
Conclusions EB is a safe technique to permit non-invasive genome-wide sequencing of bAVMs. Transcriptomic data correlates well with tissues acquired with open surgery and implicates Ras-MAPK signaling in adult bAVMs. It is also now possible to correlate gene expression with CFD flow modeling in living patients. Further development of non-invasive diagnostic techniques to molecularly stratify patients are needed to better delineate divergent natural histories of bAVMs and potentially inform treatment planning.
Disclosures E. Winkler: None. D. Wu: None. D. McCoy: None. A. Abla: None. D. Cooke: None.