medRxiv : the preprint server for health sciencesMay 18, 2026

Stronger brain scans show more fluid-filled spaces are linked to worse thinking skills in long COVID

Updated Jun 27, 2026

Abstract

Essence

In neuroPASC, higher perivascular space burden on 7 Tesla MRI was linked to worse processing speed and executive function despite no group-level PVS difference.

Evidence

This observational imaging analysis compared 36 neuroPASC adults with 33 uninfected comparators using 7 Tesla MRI PVS segmentation and cognitive testing adjusted for age, sex, intracranial volume, and multiple comparisons.

Caveat

The sample was small and cross-sectional, associations differed by cognitive measure and subgroup, and the study cannot show that PVS burden causes cognitive symptoms.

Simplified

PURPOSE: Post-acute sequalae of SARS-CoV-2 (PASC) are associated with persistent neurological symptoms (neuroPASC). Perivascular spaces (PVS) in the brain may enlarge in the context of inflammation and vascular dysfunction, reflecting impaired glymphatic clearance, and have been linked to cognitive decline. SARS-CoV-2 may disrupt the blood-brain barrier and impair glymphatic function, contributing to PVS burden. This study used 7 Tesla MRI to segment and quantify PVS in neuroPASC participants and uninfected comparators and examined associations with cognitive performance.

METHODS: Adult participants (36 neuroPASC (44.3 ± 12.7 years) and 33 comparators (38.4 ± 13.0 years)) underwent a 7 Tesla MRI scan. White matter masks of the whole brain and four lobes were segmented, and semi-automated segmentation was used to quantify PVS count and volume. All participants completed cognitive testing including Trails A and B sequencing tasks; neuroPASC participants also self-reported brain fog, fatigue, anxiety, and depression. PVS count, PVS volume, and total white matter volume (WMV) between groups were compared and associations between PVS metrics and cognitive function were assessed controlling for age, sex, and intracranial volume and corrected for multiple comparisons.

RESULTS: Among neuroPASC participants, those reporting anxiety (=0.009) and depression (=0.01) had higher WMV than those without. Greater PVS burden was associated with worse cognitive performance in PASC, particularly processing speed (Trails A) and executive function (Trails B). Specifically, processing speed was negatively associated with whole-brain PVS count (= 0.008,= 0.27), frontal PVS count (= 0.03,= 0.25), and frontal PVS volume (= 0.04,= 0.23). Trails B was also negatively associated with whole-brain PVS count (= 0.005,= 0.26). In comparators, higher PVS burden (volume and count) across multiple lobes was associated with worse semantic fluency (Animal Naming). There were no other significant associations between PVS measures and neuropsychiatric tests among participants within any of the subgroups to report. p p p-FDR R p-FDR R p-FDR R p-FDR R2 2 2 2

CONCLUSION: Although group-level differences in PVS were not observed, PVS burden was meaningfully negatively associated with cognitive performance in neuroPASC, with the strongest effects in frontal regions. These findings suggest that microvascular and glymphatic alterations may contribute to the characteristic processing speed and executive dysfunction seen in neuroPASC. Elevated WMV in those with anxiety and depression may reflect heightened inflammatory vulnerability. PVS may serve as a sensitive imaging marker of glymphatic dysfunction and neuroinflammation in neuroPASC, offering insight into the mechanisms underlying cognitive impairment and potential intervention targets.