Dynamic connectivity analysis of selective serotonin reuptake inhibitor effects, a pharmacological PET/MR study

Introduction: The mechanism of action of serotonin reuptake inhibitors (SSRIs) warrants further investigation. Dynamic connectivity analysis takes into account the variability of brain networks over time, which may be more sensitive to pharmaco logical manipulation than models assuming stationary networks [1]. Hybrid PET/MR allows relating changes in functional with molecular imaging parameters. Correlation with serotonin trans-porter (5-HTT) occupancy may provide support for the relevance of alterations in brain activity patterns produced by SSRI drug challenge [2]. This may especially aid the identification of functionally specific results derived from data-intensive explora tive analyses. Methods: 27 healthy subjects (13 female, 27.9 ± 9.2 years) underwent two PET/MR scans each with double-blind pharma cological challenge with 8 mg citalopram or placebo over 8 min. [11C]DASB was applied as bolus plus constant infusion to assess 5-HTT occupancy. Binding potentials (BPP) were calculated at equilibrium obtained 50-80 min after drug challenge using metabolite corrected plasma activity from arterial blood and cerebellar gray matter as reference region [3]. Occupancy was calculated as the mean relative difference in BPP in thalamus and striatum between scans. Resting-state functional MRI (TE/TR = 30/2440 msec, 2.1 × 2.1 × 3 mm) was acquired for 40 min starting 10 m before drug infusion. Dynamic connectivity was investigated using a sliding window approach (window length = 61 sec, step length = 2 TR) and 200 regions of interest [4]. K-means clustering was performed to differentiate connectivity states. Dwell-time and connectivity within states was compared between citalopram and placebo scans and correlated with 5-HTT occupancy. Results: Clustering revealed the best fit for 5 connectivity states. During citalopram infusion subjects spent less time (p = 0.003) and showed lower connectivity (pFWE = 0.01) in state A which is characterized by high connectivity within regions of the default mode network (DMN) and between DMN, visual and somatomotor networks (SMN). The most pronounced reductions in connectivity was seen between visual and SMN and within the SMN. After infusion, subjects spent more time in state B (p = 0.02) which shows high connectivity between basal ganglia, dorsal and visual attention networks and the SMN. Further, there was a strong negative correlation (r =-0.84, pFWE = 0.09) between occupancy and connectivity within state C which is characterized by strong connectivity within the DMN and between DMN and fronto-parietal regions and basal ganglia. This effect was especially apparent for the connections between DMN and basal ganglia and the ventral attention network. Conclusions: These results indicate that citalopram impacts brain connectivity within very few minutes after administration. Specifically, connectivity states with a high engagement of the DMN are affected, such that during citalopram infusion lower connectivity and less time are observed in state A. Of special interest, after citalopram infusion the decrease of connectivity in state C which shows the most pronounced involvement of the DMN is correlated with 5-HTToccupancy. Conversely, more time is spent in state B which is characterized by the strongest connectivity of the basal ganglia among all states. In summary, uncoupling of DMN connectivity may be a specific marker of acute SSRI effects. It will be interesting to investigate the relevance of this finding in clinical populations.