Background: A subset of individuals experience prolonged neurological and psychiatric symptoms following SARS-CoV-2 infection, a condition referred to as long COVID (LC). Limited evidence implicates ongoing neuroinflammatory processes as a driver of LC. This study investigates neuroinflammation in LC using translocator protein positron emission tomography (TSPO PET).

Methods: 14 LC, 11 healthy control (HC) and 13 multiple sclerosis (MS) participants were included in the study. They underwent [¹¹C]PK11195 TSPO PET and 3T magnetic resonance imaging (MRI) to evaluate glial activation, white matter (WM) pathology and brain volumetrics. Serum neurofilament light chain (NfL) and glial fibrillary acidic protein (GFAP) were measured as markers of neuronal and glial damage. LC participants completed neurological examinations and mental health assessments.

Results: TSPO availability, measured as distribution volume ratio (DVR), was not elevated in LC compared to HCs but was significantly lower in LC compared to MS (WM DVR 1.03 vs. 1.06; p = 0.007). Individuals imaged within 16 months of SARS-CoV-2 infection showed higher WM DVR compared to those with a longer disease duration (1.05 vs. 1.02; p = 0.04). Moreover, lower quality of life was associated with higher DVRs in the hippocampus, amygdala and thalamus (ρ = - 0.83- - 0.70), and depression and anxiety correlated positively with DVRs in the hippocampus and amygdala (ρ = 0.75-0.97).

Conclusions: LC TSPO availability did not differ from HCs in any studied brain area. However, lower WM TSPO availability in individuals with longer LC duration suggests COVID-19-associated neuroinflammation may subside with time, while the association between limbic TSPO availability and LC severity may imply a role for limbic activity in LC symptomology.