Serum Markers of Blood-Brain Barrier Remodeling and Fibrosis as Predictors of Etiology and Clinicoradiologic Outcome in Intracerebral Hemorrhage

Matthew D. Howe, Liang Zhu, Lauren H. Sansing, Nicole R. Gonzales, Louise D. McCullough, Nancy J. Edwards

Background: Intracerebral hemorrhage (ICH) is a stroke subtype associated with high disability and mortality. There is a clinical need for blood-based biomarkers that can aid in diagnosis, risk stratification, and prognostication. Given their role in the pathophysiology of ICH, we hypothesized markers of blood-brain barrier disruption and fibrosis would associate with neurologic deterioration and/or long-term functional outcomes. We also hypothesized these markers may be unique in patients with ICH due to cerebral amyloid angiopathy (CAA) vs. other etiologies. Methods: Seventy-nine patients enrolled in prospective ICH registries at two separate hospitals (the University of Texas Health Science Center at Houston and Hartford Hospital) were included in this study. We assessed initial injury severity and admission variables along with measures of inpatient deterioration (hematoma expansion, perihematomal edema (PHE), and early and delayed neurologic deterioration) and functional outcome [modified Rankin Scale (mRS) score at discharge and 90 days]. Serial biospecimens were obtained at 5 pre-specified timepoints (within 24 h, 1-2, 3-5, 6-8, and 10 days); serum samples were analyzed for fibronectin, all three TGF-β isoforms, and 7 matrix metalloproteinases (MMPs). Results: In our initial correlation analysis, MMP 10 and 3 were associated with hematoma expansion and early neurologic deterioration, whereas MMP 8 and MMP 1 were associated with PHE and delayed neurologic deterioration (respectively). Subacute levels of MMP 8 (sampled from day 6-10) positively correlated with PHE even after adjusting for multiple comparisons (p = 0.02). Acute levels of MMP 1, TGF-β1, and TGF-β3 were predictive of functional outcome, with TGF-β1 and TGF-β3 associating with 90 day mRS independent of age, hematoma volume, hemorrhage location, GCS, and IVH [p = 0.02; OR 1.03 (95% CI 1.0-1.05); p = 0.03; OR 3.1 (95% CI 1.1-8.8)]. When evaluated together as a panel, the cytokines distinguished patients with ICH due to CAA vs. ICH due to hypertension (AUC 0.81). Conclusions: Serum levels of fibronectin, TGF-β, and MMPs may be useful in refining ICH etiology and prognosis. Further large-scale studies are needed to confirm these findings, particularly regarding patients with CAA.