Is FDG-PET metabolism associated with cognitive performancein long-Covid patients?

Abstract

Background: Current evidence indicates that COVID-19 infection can lead toneurological complications that persist beyond 12 weeks of infection (long-COVID),often associated with cognitive decline. However, the underlying mechanisms remainunclear. Our aim is to study the brain metabolism of patients with long-COVID and itsassociation with cognitive performance.Method: Individuals with cognitive complaints for at least a month after COVID-19infection from an Argentine cohort of long-COVID. Their brain glucose metabolismwas assessed by FDG-PET imaging, and the cognitive symptoms were monitored usingcognitive evaluation. The FDG-PET images were normalized using the individuals’global mean value, and regions of interest mean signal was extracted using ICBM152atlas. After applying Principal Components Analysis (PCA) to reduce dimensionality,we conducted clustering using K-means with the two primary components extractedto create groups with similar metabolism patterns. With the neuropsychologicaldata, we generated seven composites covering cognitive domains based on z-scoreddata relative to the normal Argentine population. We conducted pairwise T-tests tocompare cognitive performance of metabolic clusters.Result: Forty-one subjects were recruited, 27 were female. Mean age of 55 years(±12) with an average of 15 years of education (±2.3). In PC1, key contributingregions included the right medial temporal lobe, right hippocampus and bilateralamygdala, exhibiting hypometabolism, and bilateral frontal lobe, displaying preservedmetabolism (Figure 1). PC2 was characterized by hypometabolism in bilateral frontallobe and right occipital lobe (Figure 2). Three distinct clusters were identified: Cluster 1 and 2, differentiated by PC1 and Cluster 3, distinguished from Cluster 1 and 2 byPC2 (Figure 3). Regarding cognitive assessments, we observed statistically significantdifferences between Cluster 1 and 2 in executive composite (p=0.049) and globalcomposite (p=0.025).Conclusion: Our study identified three distinct clusters based on brain metabolism,with differences in executive functions between the two of them. We consider thatFDG-PET only partially explains cognitive performance, mood and neural networks areprobably relevant contributing factors to cognition.