Association of Plasma Inflammatory Proteins and Intracerebral Hemorrhage in Pediatric Brain Arteriovenous Malformations.

Introduction: Brain arteriovenous malformations (bAVMs) are the leading cause of intracerebral hemorrhage (ICH) in children. Prior studies have implicated inflammation as a major contributor to bAVM pathogenesis and rupture, with increased inflammatory cell infiltration and interleukin protein levels reported in bAVM tissue. However, few studies have evaluated circulating inflammatory proteins as potential biomarkers for ICH in children with bAVMs. Hypothesis: Plasma levels of inflammatory markers differ in pediatric patients with ruptured bAVMs compared to unruptured bAVMs. Methods: This cross-sectional study included children with bAVMs (< 19 years old at presentation) evaluated at our institution between 2005 and 2022, who were enrolled in a prospective cohort study. Plasma samples (n = 52) were collected after diagnosis but before treatment and analyzed using a commercially available, comprehensive multiplex panel of 384 inflammatory markers. To control for confounders, such as the acute response to ICH, we performed segmented multivariable linear regression analysis, which allows initial protein levels to vary but assumes slopes to be constant within groups after a specified time point (e.g., 32 days). The model adjusted for age, sex, time to blood collection, and interaction of time with ICH status to identify plasma proteins associated with ICH. Exponentiated coefficients are presented and interpreted as proportional increases (PIs). We report nominally associated (p ≤ 0.05) and significantly associated proteins based on Bonferroni adjustment (p = 0.05/368 proteins = 0.00014). Results: We analyzed blood samples collected prior to treatment in 27 unruptured bAVMs (non-ICH cases) and 25 ruptured bAVMs. The median time from study enrollment to sample collection was 126 days (IQR = 64-247) for non-ICH cases and 11 days (IQR = 4-37) for ICH cases. We observed four different patterns of association of biomarker levels with ICH status:1) no differences between ICH and non-ICH groups over time (n=306 markers); 2) no long-term differences between ICH and non-ICH groups but initial trajectories differ after presentation (n=29); 3) long-term differences between ICH and non-ICH groups and initial trajectories differ after presentation (n=5); and 4) consistent differences between ICH and non-ICH groups over time (n=28). Thus, 33 inflammatory proteins were nominally associated (p < 0.05) with ICH either with or without initial trajectory differences (patterns 3 or 4, PI ranging from 0.71 to 4.23). Five of these 33 proteins have been associated with bAVM development and rupture in previous studies (VEGFA, MMP-10, TNF, ANGPTL2, and CSF1). No proteins were statistically significant after adjustment for multiple testing. Conclusion: We measured a large set of inflammatory proteins in blood samples from pediatric bAVM patients and identified 33 proteins with levels nominally associated with ICH with or without initial trajectory differences, including elevated levels of VEGFA and MMP-10. Further studies are needed in larger bAVM cohorts and should include serial measurements of angiogenic or inflammatory proteins to validate the association of these potential biomarkers of ICH risk in children with bAVM.