Neural correlates of aversive learning as mechanisms linking childhood trauma with psychopathology

Poster presented virtually at the 2020 OHBM conference. Abstract Introduction: Childhood trauma (CT) is common (McLaughlin, 2013) and one of the most robust and global risk factors for psychopathology (Caspi, 2014; Keyes, 2012; McLaughlin, 2010). One potential mechanism linking CT exposure to psychopathology is aversive learning, a phenomenon that is well understood in adults (Fullana, 2016; Hartley, 2013; LeDoux, 2003) and increasingly in children and adolescents (Lau, 2011). Learning is an integrative process that unfolds over time, yet many previous studies neither investigate the temporal dynamics underlying the learning process nor incorporate multiple modalities. This study aims to describe the temporal dynamics of neural activation during aversive learning and concurrent patterns of functional connectivity, as well as how CT may disrupt these neural correlates in ways that contribute to psychopathology. Methods: 147 children, aged 8-16 years, with (n=77) and without (n=70) exposure to CT underwent a block-design differential fear conditioning procedure during an fMRI scan. Images were acquired using a 3T Phillips Achieva scanner with a 32-channel head coil. BOLD signal during the functional run was recorded using a gradient-echo T2*-weighted echo planar image (EPI) sequence. Thirty-seven 3-mm-thick slices were acquired sequentially and parallel to the AC-PC line (TR=2 s, TE=25 ms, flip angle=79°, FOV=224x224x132.6, voxel size=3 mm3, inter-slice gap=0.6 mm, matrix size=76x74). Dynamic patterns of learning were examined in voxel-wise parametric modulation analyses and region-of-interest analyses, and functional connectivity was assessed with generalized psychophysiological interaction analyses. Multiple regression was used to examine associations with psychopathology symptoms collected at baseline, as well as at two-year follow-up. Nonparametric mediation models with 10,000 simulations were run to test whether the relationship between CT and psychopathology at baseline and follow-up was mediated by measures of activation and connectivity during fear conditioning. Results: In children, canonical salience network regions (including amygdala, insula, anterior cingulate cortex) were active to the CS+ relative to CS-, and exhibited habituation across learning blocks (decreasing activation to CS+>CS-). Default mode network regions (including hippocampus, frontal pole, vmPFC, and posterior cingulate cortex) were active to CS- relative to CS+, and exhibited safety-signaling across learning blocks (increasing activation to CS->CS+). Children with CT displayed blunted habituation slopes to CS+>CS- in right amygdala (F(1, 1023)=7.93, p=.005, Figure 1) and blunted safety-signaling slopes to CS->CS+ in right hippocampus (F(1,1023)=5.38, p=.02) and frontal pole (F(1, 878)=4.5, p=.03). Additionally, children with CT showed elevated functional connectivity of amygdala with fronto-parietal regions associated with attention direction and initiation of defensive responses to CS+>CS-, and reduced amygdala-hippocampal connectivity (Figure 2). Patterns of altered dynamic neural response and task-based connectivity were associated with transdiagnostic psychopathology (depression, panic, generalized anxiety, externalizing, and PTSD) at baseline and follow-up. Critically, the pathway between CT and externalizing at two-year follow-up was mediated by blunted patterns of activation in right amygdala (p=.004, 95% CI .61-3.34) and insula (p=.05, 95% CI .08-2.59), controlling for baseline symptoms. Reduced amygdala-hippocampal connectivity mediated the pathway between CT and panic (p=.003, 95% CI .25-1.54) and generalized anxiety (p=.04, 95% CI .08-1.5). Conclusions: Alterations in aversive learning processes and the dynamic communication between salience network and default mode network regions may be a key mechanism underlying the link between CT and psychopathology. Disruptions in neural activation versus connectivity patterns may set some children on trajectories towards differentiable clinical symptom profiles