Childhood trauma-related brain alterations in bipolar disorder: an ENIGMA-bipolar disorder study

<b>Introduction</b>: Childhood trauma is experienced by more than 50% of people with severe psychiatric disorders and is a risk factor for bipolar disorder (BD). Previous studies have investigated associations between childhood trauma, brain morphology, and the likelihood of developing BD. However, most prior studies have been conducted in small samples using heterogeneous methodologies (e.g., different brain imaging methods or childhood trauma assessments), resulting in inconsistent findings. Our study from the Enhancing NeuroImaging Genetics through Meta-Analysis Bipolar Disorder working group (ENIGMA-BD) aims to determine the impact of childhood trauma on brain morphology in BD, using the largest international sample of individuals with bipolar disorder assembled to date.<br/><b><br/>Methods</b>: Nineteen independent cross-sectional samples, including 1105 BD cases (56% females) and 2344 healthy controls (HCs; 57% females) were collated via ENIGMA-BD. Severity of childhood trauma was measured using the Childhood Trauma Questionnaire (CTQ) at all sites. Structural 3D T1-weighted brain MRI scans were processed using ENIGMA-standard FreeSurfer and quality control to derive measures of subcortical volumes (16 regions), cortical thickness and surface area (68 regions from the Desikan-Killiany atlas). Data were harmonised for potential site differences using ComBat. Linear regressions were used to determine the associations between BD diagnosis (BD versus HC), severity of childhood trauma (CTQ total score) and their interaction on brain morphology. Age, sex, and total intracranial volumes (for subcortical volume analyses only) were included as covariates. False-Discovery Rate correction (q&lt;0.05) was applied to account for the number of regions included (separate analyses for subcortical volumes, cortical thickness and surface area).<br/><b><br/>Results</b>: Fifty-three percent of BD cases (N=584) and 24% of HCs (N=560) reported moderate-to-extreme levels of exposure to at least one type of trauma measured by the CTQ. A diagnosis of BD was significantly associated with larger bilateral lateral ventricles and smaller bilateral hippocampal volumes (all q&lt;0.011) and widespread patterns of thinner cortex (56 regions out of 68; all q&lt;0.05). Independently of diagnosis, the severity of childhood trauma was associated with smaller surface areas of the bilateral fusiform, medial orbitofrontal, rostral middle frontal gyri, as well as smaller surface of the left lateral orbitofrontal, superior temporal, pars orbitalis, rostral anterior cingulate gyri, and right inferior temporal, parahippocampal, precentral, superior frontal and supramarginal gyri (all q&lt;0.047). The diagnosis-by-trauma interaction was not significantly associated with brain morphology. Investigation of the trauma subdomains revealed a significant relationship between diagnosis and severity of sexual abuse on surface area of the right entorhinal gyrus. A subsequent moderation analysis indicated that the severity of sexual abuse was significantly associated with smaller surface of the entorhinal gyrus in HCs (b=-1.401, p=0.021), and larger surface of the entorhinal gyrus in BD (b=3.257, p=0.002).<br/><b><br/>Conclusions</b>: In the largest analysis of childhood trauma and brain morphology in BD to date, BD was associated with a widespread pattern of thinner cortex and smaller ventricle/hippocampal volumes, while childhood trauma was associated with smaller surface area of regions involved in networks critical for emotional and cognitive processes. Sexual abuse in particular, may contribute to maladaptive mechanisms in BD. Future functional studies are underway to confirm this interpretation.