Resting-state functional connections with the hippocampus and with the caudate nucleus predict working memory performance in multiple sclerosis.

Working memory (WM) dysfunction is common in multiple sclerosis (MS). WM deficits also contribute to episodic memory (EM) dysfunction, a prevalent and disabling cognitive symptom of the disease. Functional connectivity between the hippocampus and caudate nucleus supports EM, but its specific influences on WM in MS are unknown. Resting-state functional connectivity (RSFC) can identify spontaneous, baseline connectivity patterns that predict cognitive impairment, thus offering clinical utility. The present study examined whether hippocampal and caudate RSFC could predict WM performance in people with MS. Across three independent studies (N = 78; 42 MS, 36 healthy individuals (HC)), seed-to-voxel RSFC between the hippocampus, caudate nucleus, and other regions across the brain was quantified. These connections' predictive influence on WM performance was then examined using a global WM performance measure for a subset with available neuropsychological data (26 MS, 15 HC). MS participants displayed stronger connectivity between the right hippocampus and left dorsomedial prefrontal cortex, compared to HC participants. Within MS participants, stronger connectivity between the left hippocampus (LHipp) and left ventral anterior cingulate (LvACC), and between the left caudate (LCaud) and right insula, were observed. Stronger dysconnectivity between the LHipp and left supramarginal gyrus (LSMG) also emerged. Stronger LHipp-LvACC connectivity predicted worse WM performance, whereas stronger LHipp-LSMG connectivity and LCaud-RInsula connectivity each predicted better performance. The hippocampal connections were also inversely correlated. Findings identify caudate and hippocampal circuits whose aberrant, baseline connectivity could serve as neural markers for impending WM dysfunction in MS.