Altered cerebrovascular reactivity along functional networks in older adults with long COVID

Background Cerebrovascular reactivity (CVR), an indicator of neurovascular function associated with cognitive decline, relies on vascular and inflammatory processes, making it a potential mechanism of interest driving neurological changes in long COVID among older adults. Given the variability in symptoms associated with long COVID, distribution corrected z‐scores (DisCo‐Z), a subject‐specific abnormalities (SSA) approach more suited for heterogenous conditions, was employed to examine CVR and long COVID in older adults. Relationships between CVR, post‐COVID subjective cognitive decline (pcSCD), and cognitive function were examined. Method 15 older adults with long COVID (M age = 63.9, 6 males) and 24 healthy older adults (M age = 66.7, 6 males) performed a breath‐holding task while in the MR scanner to examine CVR. DisCo‐Z was used to examine the presence and size of extreme clusters of increased or decreased CVR (>100 voxels) within the 7 Yeo functional networks. Non‐parametric tests were used to examine group differences in presence and mean size of extreme clusters. Finally, relationships between: 1) extent of pcSCD and extreme cluster values in long COVID, and 2) cognitive testing scores and global “burden” (cluster size mean and sum) were examined. Result Groups did not differ significantly on gender (p = .323) or age (p = .324). Greater incidence of positive extreme clusters within Ventral Attention (p = .028) and Limbic (p = .049) networks were observed in long COVID (with minimal incidences of negative extreme clusters across networks). Group differences were observed for mean cluster size in the Ventral Attention network (p = .016), as well. Extent of pcSCD was positively correlated with DisCo‐Z values for the Ventral Attention network (p = .021) with a trending level of significance for the Limbic network (p = 0.053). Finally, within the full sample, global CVR burden was negatively correlated with memory test scores (p = 0.014). Conclusion Using the SSA approach, group differences in CVR were observed in older adults experiencing long COVID. In particular, differences were observed within functional networks thought to drive attention, executive function, and emotional control, which was correlated with pcSCD. Finally, global CVR burden was negatively correlated with memory. These findings, though preliminary, suggest CVR may represent a potential neurobiological mechanism driving pcSCD in older adults.