Functional brain networks related to processing speed and memory in SLE: a connectome-based modelling study.

OBJECTIVES

Cognitive dysfunction is a common neuropsychiatric manifestation in SLE, particularly affecting processing speed (PS) and memory. This study aims to identify behaviourally relevant topological networks of functional connectivity underlying neuropsychological test performances, using connectome-based predictive modelling (CPM).

METHODS

Forty-three SLE patients and 34 controls underwent overall cognitive screening, with SLE patients additionally undergoing neurocognitive assessments for PS and memory. Resting-state fMRI was performed to construct functional connectivity matrices. CPM with leave-one-out cross-validation was employed to model the association between functional connectivity and cognitive performances. Linear regression and moderation analyses were employed to identify risk factors for cognitive dysfunction in SLE.

RESULTS

CPM models predicted PS and memory scores in SLE in cross-validation, with correlation coefficients (r) between predicted and observed scores ranging from 0.40 to 0.42 (all P < 0.01). Key brain regions contributing to these models commonly included the cingulate cortex, dorsolateral prefrontal cortex, hippocampus, thalamus and cerebellum, which were critical nodes of default mode, frontoparietal and subcortical networks. A combined CPM model incorporating these functional connectomes predicted overall cognitive ability in SLE (r = 0.38, P < 0.05), with no predictive power in controls, confirming specificity. Additionally, higher damage indices were associated with slower PS. Disease activity, daily glucocorticoids dosage and anticardiolipin antibodies levels moderated associations between functional connectivity strength and cognitive performances.

CONCLUSION

CPM identified brain networks that underlie critical cognitive functions in individuals and these neural fingerprints could potentially assist the diagnosis of cognitive dysfunction in SLE.