Auditory hallucinations, childhood sexual abuse, and limbic gray matter volume in a transdiagnostic sample of people with psychosis

The study addresses whether childhood sexual abuse (CSA) is a specific risk factor for auditory hallucinations (AH) in psychotic disorders, how sex moderates this relationship, and whether CSA is associated with limbic gray matter volume (GMV) differences (amygdala and hippocampus). Given the heterogeneity of psychotic disorders and the clinical impact and treatment refractoriness of AH, identifying homogeneous subgroups is important. Prior work links childhood maltreatment to psychosis and specifically implicates CSA as a potent predictor of AH. Females are disproportionately exposed to CSA, potentially increasing their risk for AH. Neurodevelopmental impacts of maltreatment often involve limbic regions (amygdala, hippocampus), which are also implicated in AH pathophysiology (emotion, memory, and language network interactions). The study tests whether CSA is elevated in AH vs. NAH patients and healthy controls, whether this effect is stronger in females, and whether CSA severity relates to amygdala/hippocampal GMV.

Existing literature shows that childhood maltreatment increases risk for psychosis and AH, with several studies suggesting CSA may be a particularly strong predictor of AH across clinical and nonclinical populations. Limbic structures, notably the amygdala and hippocampus, are frequently implicated in maltreatment-related neurodevelopmental changes. While hippocampal volume reductions are relatively consistent, amygdala findings are mixed (both increases and decreases reported), potentially due to differences in populations, co-occurring maltreatment types, and medication effects. AH involve disrupted memory, learning, and emotion systems implicating limbic circuits; functional imaging often shows altered limbic activity and connectivity during AH or emotion processing, but structural correlates can indicate trait abnormalities potentially arising from maltreatment. Some structural MRI studies associate limbic alterations with AH, suggesting a path by which CSA could elevate AH risk.

Design: Cross-sectional neuroimaging study in a transdiagnostic psychosis sample with healthy controls. Participants: 118 recruited (patients aged 18–50 with schizophrenia, schizoaffective disorder, or bipolar disorder with psychotic features; controls without psychiatric illness/psychosis or family history). Exclusions: non-English fluency, hearing impairment, significant neurological/medical illness, MRI contraindications, ECT within past year, recent substance abuse/dependence per DSM-IV-TR thresholds. Data collection: Nov 2014–Mar 2019 at McLean Hospital; most completed all procedures in one visit. Diagnostic and clinical assessment: Primary diagnosis via SCID for DSM-IV-TR. Lifetime AH determined via SCID item B16 (score=3 indicates AH). Current symptoms measured with PSYRATS-AH, SAPS (AH items excluded for overall positive symptom score in analyses), SANS, YMRS, and MADRS. Maltreatment: Childhood Trauma Questionnaire (CTQ) measuring sexual abuse, physical abuse, emotional abuse, physical neglect, emotional neglect, plus minimization/denial scale. MRI acquisition: Siemens TIM Trio 3T, 32-channel head coil; T1-weighted multi-echo MPRAGE (TE1/TE2/TR 3.31/6.99/2530 ms; TI 1100 ms; flip angle 7°; FOV 256 mm; 256×256; 176 slices; 1.0 mm isotropic; GRAPPA=2; 6:03 min). Image analysis: FreeSurfer 5.3 automated pipeline with quality control, skull stripping, Talairach transform, subcortical segmentation, intensity normalization, surface reconstruction, and Desikan parcellation. Regions of interest: bilateral amygdala and hippocampus GMV. Intracranial volume (ICV) estimated; GMV values normalized by ICV (GMV/ICV). Statistical analyses: Group comparisons via ANOVA/chi-square; post hoc t-tests. Primary analyses tested: (1) group differences in CSA (AH, NAH, HC) via one-way ANOVA; (2) moderation by sex via ANOVA (group × sex); (3) specificity across non-CSA CTQ domains via one-way ANOVAs; and ANCOVA controlling for sex, DSM diagnostic class (schizophrenia spectrum vs bipolar), chlorpromazine (CPZ) equivalents, and other CTQ domains. Minimization/denial rates compared via chi-square. Correlations assessed CSA with current AH severity (PSYRATS-AH) and with other symptom domains (SAPS excluding AH items, SANS, YMRS, MADRS). Supplementary regressions examined sex × CSA interaction and covariate-adjusted CSA–AH severity. Neuroimaging: Pearson correlations between CSA and GMV (left/right amygdala/hippocampus) in patients; supplementary regressions tested sex moderation and covariate-adjusted CSA–GMV associations (including AH status, DSM class, CPZ equivalents, and other maltreatment domains). Missing data handled pairwise; 112 participants had usable GMV after quality control.

- Sample: 118 participants scanned; 112 had usable GMV. Clinical groups: AH n=41, NAH n=37, HC n=37; groups largely matched except patients had fewer years of education than controls and AH had higher CPZ equivalents than NAH. - Maltreatment across groups (CTQ): Both AH and NAH reported higher levels than controls across all five domains. CSA specifically was higher in AH vs NAH and vs controls. Table 2: Sexual abuse ANOVA F=7.82 (p<.01); AH vs HC t=3.90 (p<.001); NAH vs HC t=2.03 (p<.05); AH vs NAH t=2.21 (p<.05). Other domains (physical abuse, emotional abuse, emotional neglect, physical neglect) showed patient > control but no AH vs NAH differences. - Sex moderation: CSA differences were driven by females with AH (sex × AH status interaction F=4.91, p=0.009). - Robustness: Among patients, ANCOVA controlling for sex, DSM diagnostic class, CPZ equivalents, and other maltreatment types showed CSA remained associated with AH status (F=3.88, p=0.02). No group differences in CTQ minimization/denial. - Symptom correlations: CSA positively correlated with current AH severity (PSYRATS-AH) r=0.26, p=0.03; not with overall positive (excluding AH items) r=0.17 (p=0.16), negative r=−0.01 (p=0.97), manic r=0.16 (p=0.17), or depressive symptoms r=0.06 (p=0.63). Regression controlling sex, DSM class, CPZ, and other maltreatment: CSA remained associated with AH severity (b=0.77 [SE 0.32], t(69)=2.41, p=0.02). - Neuroimaging: In patients, CSA correlated with larger left amygdala GMV (r=0.30, p=0.01), not right amygdala (r=0.12, p=0.33), nor left hippocampus (r=0.06, p=0.65) or right hippocampus (r=0.18, p=0.14). No sex × CSA interaction for left amygdala GMV. Covariate-adjusted regression (controlling AH status, DSM class, CPZ equivalents, other maltreatment): CSA remained positively associated with left amygdala GMV (b=1.69×10^-5, t(71)=2.52, p=0.01). - Overall, findings identify a subgroup (females with AH) with heightened CSA exposure, and link CSA severity to larger left amygdala volume, independent of diagnosis, medication, and other maltreatment.

The study demonstrates a specific association between CSA and AH within psychotic disorders, beyond general elevations in maltreatment exposure. CSA severity is tied to both lifetime AH status and current AH symptom severity, with the effect most pronounced among females, suggesting a potential subgroup at elevated risk. Structurally, CSA severity relates to greater left amygdala GMV, aligning with some prior human and animal literature indicating stress-related dendritic hypertrophy in the amygdala, and implying stress-sensitive limbic changes may contribute to AH via altered emotion-memory processes and salience networks. No hippocampal GMV association was detected, underscoring the complexity of maltreatment effects across limbic structures and possible moderating factors (timing, chronicity, development, genetics, social buffering). Lateralized effects (left amygdala) may reflect power limitations or true asymmetry alterations, consistent with literature on stress impacting normative brain asymmetries and the left amygdala’s role in sustained/negative affective processing. These results support models integrating environmental adversity (CSA), sex-related risk, and limbic structural alterations in the pathophysiology of AH.

CSA appears specifically elevated among psychosis patients with AH, particularly females, and CSA severity correlates with current AH severity. CSA is also associated with larger left amygdala GMV, independent of diagnosis, medication, and other maltreatment exposures, suggesting a potential neurodevelopmental pathway linking CSA to AH. The findings advocate for considering maltreatment history and sex when assessing and treating psychosis. Future research should use prospective, longitudinal designs in high-risk cohorts, incorporate multimodal neuroimaging (including functional measures), probe amygdala subregions, and integrate multi-source maltreatment assessments to clarify causal mechanisms and identify targets for personalized interventions.

- Modest sample size; insufficient power to test AH vs NAH differences within specific diagnoses (schizophrenia, schizoaffective, bipolar). - Potential influences of race, culture, and gender identity were not fully examined due to sample size and limited gender identity data. - AH group had higher CPZ equivalents than NAH; although controlled statistically, medication effects cannot be fully excluded. - Focus on GMV precluded assessment of functional alterations related to CSA during emotion or learning processes. - Cross-sectional design prevents causal inferences about timing and directionality; CSA measured via retrospective self-report with potential underreporting, especially among males, despite evidence of reliability and no detected differential reporting biases. - Lack of detailed data on timing/chronicity of CSA, developmental stage, genetic moderators, and protective factors that might modify brain-psychosis relationships.