AIDS. 2026 Jul 3. doi: 10.1097/QAD.0000000000004573. Online ahead of print.
ABSTRACT
OBJECTIVES: Carotid atherosclerosis is an established risk factor for cognitive impairment. FDG-PET detects subtle inflammatory changes in the arterial wall that may precede plaque formation and rupture. We investigated whether carotid inflammation is associated with cognitive function and its relative importance as a predictor of concurrent cognitive function compared with other factors.
DESIGN/METHODS: : In this cross-sectional study, we recruited individuals ≥40 years with undetectable viral load at moderate-to-high cardiovascular risk. Participants underwent FDG-PET, stress and cognitive assessments. We used pairwise partial correlations to identify significant associations between arterial inflammation and cognition. We applied a machine learning analysis workflow to evaluate the ability of carotid inflammation alone, and in combination with other factors, to predict cognition.
RESULTS: Among 47 people, carotid inflammation was independently correlated with global cognition [r = -0.405, 95% confidence interval (CI) -0.618 to -0.100, P = 0.007]. In the machine learning analysis, the probability of carotid inflammation predicting cognitive function within one standard deviation of the true value was 68% (95% CI 52-80). The predictive accuracy improved marginally when individual cardiometabolic factors were added to the model (73%, 95% CI 57-84). When all potential variables were considered, predictive accuracy improved substantially (85%, 95% CI 71-93), with carotid inflammation remaining as the most important cardiometabolic predictor of cognitive function.
CONCLUSION: Carotid inflammation on FDG-PET outperformed most other factors in predicting concurrent cognitive function and provided unique predictive information beyond traditional cardiometabolic risk factors. Future longitudinal studies should examine whether carotid inflammation predicts incident cognitive impairment, which would support its use as a clinical risk marker in people with HIV.
PMID:42397671 | DOI:10.1097/QAD.0000000000004573