Biomarkers.

Background: Long-COVID is characterized by persistent symptoms post-infection with SARS-CoV-2. This condition includes neurological manifestations and has been proposed as a potential risk factor for the development of dementia. Individuals presenting with dementia due to Alzheimer's disease have dysfunctional brain metabolism, including metabolic brain network (MBN) hypoconnectivity. However, whether long-COVID alters brain metabolic architecture remains elusive. Here, we aimed to evaluate the brain metabolic connectivity in a Brazilian cohort of individuals presenting with long-COVID.

Method: [F]FDG-PET images were acquired from 52 community-dwelling Brazilians above 50 year old. Standardized uptake value ratio (SUVr) parametric maps were processed to a common 8 mm FWHM and generated using the pons as the reference region (Figure 1). We extracted the mean values of regions of interest using the ICBM152 atlas. [F]FDG-PET MBNs were constructed using a novel multiple sampling scheme, which assembles a stable group representative MBN based on bootstrap (n = 2000). Adaptive Synthetic Sampling Approach for Imbalance (ADASYN) was used to account for group imbalance and generated the ADA-MBNs. Graph measures, including density, global efficiency, average degree, and assortativity coefficient were computed. Data were corrected for multiple comparisons using the False Discovery Rate (FDR) method (p<0.05).

Result: 41 individuals with long-COVID and 11 healthy controls (HC) were included (Table 1). We observed that long-COVID individuals present PET hyperconnectivity in both MBN and ADA-MBN. (Figure 2a-b). The long-COVID group presented increased density, global efficiency and average degree whereas assortativity coefficient were reduced in both MBN and ADA-MBN.

Conclusion: Our findings showed that individuals with long-COVID presented a brain metabolic hyperconnectivity, which is supported by increased density and average degree and may indicate a potential compensatory mechanism within the brain. In addition, the increase in global efficiency indicates that the brain of long-COVID individuals exchanges metabolic information more efficiently, but the decreased assortativity coefficient suggests vertices with different properties connect to each other. Further longitudinal studies should follow these individuals for assessing microstructural and cognitive changes.