Purpose: Most isolated congenital vascular malformations are genetically characterized by somatic variants in various genes. However, large clinical variability challenges the analysis of correlations between genotype and phenotype. In sporadic extracranial arteriovenous malformation (AVM), mosaicactivating variants in the RAS/MAPK PTEN signaling pathways have been described. Our aim was to distinguish AVM phenotypes based on clinical and angiographic findings in relation to corresponding pathogenic variants. Methods: Within our prospective vascular malformation cohort, that exists since 2018 and is continuously enrolling consecutive patients, we reviewed patients with angiographically verified AVM in whom a genetic study of the affected tissue using high-throughput sequencing by the TruSight Oncology500 panel had been performed. Each patient's initial clinical evaluation, imaging and histopathologic analysis were assessed concerning patient and family history, symptoms, longitudinal overgrowth, soft tissue hypertrophy, angiographic classification and histology. Results: Overall, 37 patients (18 males and 19 females) were included for analysis. Most of the AVMs (73%) were located in the extremities. We detected known variants in MAP2K1 (n=9), KRAS (n=7), RASA1 (n=1), but also variants in other genes of the RAS pathway (RIT1, PTPN11). We also identified alterations in the PI3K/AKT/mTOR signaling pathway with a PIK3CA variant (n=2), and in the PTENPathway with a combination of somatic variants in PTEN and GNAQ (n=1), not yet associated with extracranial AVM. Additional variants of unknown significance were detected in all patients. Comparing AVM with KRAS variants to AVM with MAP2K1 variants, AVM with KRAS variants were more diffuse, involving whole extremities (71% vs 22%) and were frequently associated with ulcerations (71% vs 11%) and longitudinal overgrowth (86% vs 22%). Conclusion: Though a larger number of cases would be required to draw definite conclusions, our study contributes to characterizing the molecular landscape of extracranial AVMs and points to a genotypephenotype correlation regarding variants in MAP2K1 and KRAS.
