Bloodstream infections and Blood-Brain barrier Permeability: An observational cohort study

Underwood, Jonathan, Davies, Kate ORCID: https://orcid.org/0000-0002-9807-1231, Loveless, Sam ORCID: https://orcid.org/0000-0002-5124-4115, Cercignani, Mara ORCID: https://orcid.org/0000-0002-4550-2456, Dowell, Nicholas G., Periche-Tomas, Eva, Ronen, Itamar, Evans, John, McLaren, James E. ORCID: https://orcid.org/0000-0002-7021-5934 and Harrison, Neil A. ORCID: https://orcid.org/0000-0002-9584-3769 2026. Bloodstream infections and Blood-Brain barrier Permeability: An observational cohort study. Brain, Behavior, and Immunity 135 , 106518. 10.1016/j.bbi.2026.106518

PDF - Published Version Available under License Creative Commons Attribution. Download (3MB)

Abstract

Background Sepsis associated encephalopathy is a poorly understood yet common complication of sepsis. Immune challenge and animal models implicate blood–brain barrier (BBB) dysfunction and microglial activation in the hippocampus as key drivers of pathology. In this exploratory study, we used bloodstream infections (BSI) as a model of severe infection to investigate the relationship between changes in cognition with plasma biomarkers of brain injury, BBB permeability and microglial activation quantified using advanced neuroimaging. Methods In this longitudinal exploratory study we recruited patients hospitalised with BSI as well as age matched controls. Cognition was assessed at the bedside and at convalescense using a CANTAB battery. MRI scanning was performed as soon as practicable after hospital discharge. Hippocampal and thalamic BBB permeability was assessed using dynamic enhanced contrast MRI (DCE-MRI). Microglial morphometry was assessed using diffusion-weighted magnetic resonance spectroscopy (DW-MRS) of the left thalamus. Plasma glial fibrillary acidic protein (GFAP) and neurofilament light protein (NFL) were quantified using Single molecule array (SiMoA) technology. Results 21 patients with BSI and 16 age matched controls were enrolled (median [IQR] 64.3 [51.1–68.5] years). Patients with BSI had poorer cognitive function, particularly attention (RVP A-prime z-score −1.08 [95% CI −1.75, −0.61], p < 0.001). This deficit correlated with the magnitude of the inflammatory response quantified by peak c-reactive protein (CRP). Furthermore, patients with BSI had higher concentrations of NFL, which was correlated with poorer cognitive function and more severe illness. At the scanning visit (median: 32 days from discharge) cognition had largely normalised and was similar to controls. However, patients with BSI had significantly more depressive symptoms (p < 0.01). There were no differences in BBB permeability or metabolite apparent diffusion coefficients between patients with BSI and controls. Conclusions Our findings suggest that BSI are associated with acute, inflammation-mediated brain injury. However, cognitive dysfunction, particularly in the domain of attention, is temporary and not associated with changes in BBB permeability or microglial activation at convalescence. Given persistent mood dysfunction and elevated biomarkers of neuronal injury longitudinal study is warranted to determine long-term sequelae of BSI and other severe infections.

Item Type:	Article
Date Type:	Publication
Status:	Published
Schools:	Schools > Medicine Research Institutes & Centres > Cardiff University Brain Research Imaging Centre (CUBRIC)
Publisher:	Elsevier BV
ISSN:	0889-1591
Date of First Compliant Deposit:	2 March 2026
Date of Acceptance:	21 February 2026
Last Modified:	02 Mar 2026 12:45
URI:	https://orca.cardiff.ac.uk/id/eprint/185371

Actions (repository staff only)

Edit Item