Psychotic disorders exhibit significant heterogeneity in neurobiology, risk factors, symptoms, and functional outcomes. This variability poses challenges to understanding their pathophysiology and hinders the development of effective treatments. Auditory hallucinations (AH), a prominent symptom in many psychotic disorders (affecting a substantial proportion of individuals with schizophrenia, schizoaffective disorder, and psychotic bipolar disorder), have emerged as a promising avenue for identifying more homogenous subgroups. AH are often distressing and associated with poorer functional outcomes, increased relapse risk, and heightened suicide risk. However, their treatment resistance in a significant portion of cases underscores the need for a better understanding of their underlying mechanisms. Childhood maltreatment, particularly childhood sexual abuse (CSA), has been identified as a potential risk factor for AH and psychosis. While some studies suggest that different types of maltreatment do not differentially increase psychosis risk, others highlight CSA as a particularly strong predictor of AH. This association has been observed across various populations, including adults with psychotic disorders, young people at clinical high risk for psychosis, non-psychotic voice-hearers, and the general population. Given the higher prevalence of CSA victimization in females, researchers hypothesized that female victims of CSA may be at an elevated risk for developing AH. Maltreatment, including CSA, can have profound neurodevelopmental consequences, impacting brain structure and function. Meta-analyses consistently implicate the amygdala and hippocampus—limbic system structures crucial for processing emotions and episodic memory—as regions affected by maltreatment. The inconsistent findings regarding amygdala volume changes (both reduced and increased volumes reported) may be attributed to factors such as study population, co-occurrence of multiple maltreatment dimensions, and medication effects. The neural substrates of AH overlap with those associated with maltreatment, suggesting a potential pathway. Theories on AH pathophysiology highlight disruptions in memory, learning, and emotion systems, implicating the hippocampus and amygdala. Studies have reported altered limbic activity in individuals with psychosis during AH experiences, emotion processing, and responses to AH-like stimuli. Aberrant functional connectivity between limbic and cortical language areas also suggests a potential interplay of stress-sensitive networks and language network disruptions in the development of AH. While functional MRI studies have examined regional activity and connectivity, structural MRI analyses of brain structure, such as gray matter volume (GMV), offer insights into trait-level abnormalities associated with AH, potentially resulting from childhood maltreatment. This study aims to examine the interrelations between CSA, AH, sex, and GMV of the amygdala and hippocampus in individuals with psychotic disorders.

A considerable body of research links childhood maltreatment, specifically CSA, to an increased risk of developing AH and psychosis. Studies have consistently shown a correlation between CSA and AH across diverse populations, including adults with psychotic disorders, individuals at high risk for psychosis, and even non-clinical voice hearers. This suggests CSA may be a particularly potent predictor of AH. However, the mechanisms underlying this relationship remain unclear. Research also highlights the impact of maltreatment on brain structure and function, particularly affecting limbic system regions such as the amygdala and hippocampus. Studies have reported inconsistent findings on amygdala volume alterations following maltreatment, with some showing smaller and others larger volumes. This inconsistency necessitates a deeper understanding of the diverse effects of maltreatment and the involvement of other factors, like sex, diagnosis and medication. Furthermore, the neural substrates of AH show overlap with brain areas affected by maltreatment, suggesting potential common pathways. Therefore, this study aimed to bridge the gap by examining the interrelationship of CSA, AH, sex and limbic GMV in a transdiagnostic psychotic disorder sample.

This study involved a transdiagnostic sample of 118 adults (41 with AH, 37 without AH [NAH], and 37 healthy controls) aged 18-50. Participants were recruited from inpatient and outpatient services at McLean Hospital. The inclusion criteria included individuals with schizophrenia, schizoaffective disorder, or bipolar disorder with psychotic features. Healthy controls were excluded for any Axis I psychiatric disorder, history of psychosis, or first-degree relatives with psychotic disorders. Exclusion criteria for all groups included non-English fluency, hearing impairment, significant neurological or medical conditions, MRI contraindications, recent electroconvulsive treatment, and substance abuse/dependence. Primary diagnosis was determined using the Structured Clinical Interview for DSM-IV-TR (SCID). Lifetime AH was determined using the SCID item assessing the experience of hearing things others couldn't. Current symptom severity was measured using the PSYRATS-AH, SAPS (excluding AH items), SANS, YMRS, and MADRS. Participants completed the Childhood Trauma Questionnaire (CTQ) to assess experiences of CSA, physical abuse, emotional abuse, physical neglect, and emotional neglect. Sex was self-reported. High-resolution T1-weighted structural MRI data were acquired using a Siemens TIM Trio 3-Tesla MRI scanner. FreeSurfer 5.3 was used for automated segmentation of the MRI data. This included quality control, brain masking, non-brain tissue removal, Talairach transformation, segmentation of subcortical structures, and surface-based registration with the Desikan parcellation. GMV of the amygdala and hippocampus (left and right hemispheres) was extracted and adjusted for intracranial volume. Statistical analyses included chi-square tests and ANOVAs to compare groups on demographic and clinical variables. The primary analyses involved ANOVAs and ANCOVAs to compare CSA scores across groups (AH, NAH, controls), examining the moderating effect of sex, and assessing correlations between CSA scores, AH severity (PSYRATS-AH), and limbic GMV. Linear regression analyses examined the relationship between CSA and AH severity and between CSA and amygdala/hippocampal GMV, controlling for various covariates including sex, diagnosis, medication (CPZ equivalents), and other maltreatment types.

The study yielded several key findings: 1. **Higher CSA in AH group:** Individuals with psychotic disorders and a history of AH reported significantly greater CSA than those without AH and healthy controls. This difference remained significant even after controlling for sex, diagnosis, medication, and other types of maltreatment. 2. **Sex Moderation:** The association between CSA and AH was particularly pronounced among females with AH. Females with AH reported the highest levels of CSA across all groups and sexes. 3. **CSA and AH Severity:** Greater CSA severity was positively correlated with the current severity of AH symptoms, suggesting a direct link between the lifetime experience of CSA and the current intensity of auditory hallucinations. 4. **CSA and Amygdala GMV:** Higher CSA scores were significantly positively correlated with GMV of the left amygdala in the patient group. This relationship remained significant even when controlling for AH status, diagnosis, medication, and other maltreatment types. No significant association was observed between CSA and hippocampal GMV. 5. **No significant differences across diagnoses within the patient groups:** While a transdiagnostic sample was used, a notable limitation was that the sample size prevented an in depth analysis of the relationship between CSA, AH, and sex within specific diagnoses (schizophrenia, schizoaffective disorder, and bipolar disorder). The results did however hold when controlling for DSM diagnostic class, suggesting that the findings may generalize to individuals across these disorders.

The findings support prior research indicating a strong association between CSA and AH in individuals with psychotic disorders. The study extends these findings by demonstrating that this association is particularly prominent among females with AH. This suggests that females may be more vulnerable to developing AH following CSA exposure, possibly due to biological or social factors warranting further investigation. The positive correlation between CSA and left amygdala GMV in patients is noteworthy, particularly in light of the general association between psychotic illness and antipsychotic medications with amygdala volume reduction. This finding, counterintuitive in that larger amygdala volume is associated with greater CSA severity, aligns with some prior research and animal models suggesting that chronic stress can lead to dendritic hypertrophy in the amygdala, resulting in volume enlargement. This observation warrants further research to investigate specific amygdala subregions and their involvement in this complex relationship. The absence of a correlation between CSA and hippocampal GMV contrasts with some studies reporting hippocampal volume reduction in maltreatment. The complex impact of maltreatment on limbic structure is likely influenced by multiple factors, including the timing and chronicity of exposure, pubertal development, genetics, and protective factors. Future research should explore these factors and longitudinal studies are needed to clarify the temporal relationship between CSA, brain structural changes, and the development of AH.

This study confirms the strong association between CSA and AH in individuals with psychotic disorders, particularly among females. The unexpected finding of a positive correlation between CSA and left amygdala GMV warrants further investigation into the complex neurobiological mechanisms underlying this relationship. Future research should focus on longitudinal studies employing multiple measures of maltreatment and integrating brain structure and function to better elucidate the pathways linking CSA, amygdala alterations, and AH. Specific research directions should include investigation of the sex-specific vulnerability to AH in relation to CSA and examining the role of genetic and environmental factors and potential protective factors. Further, investigation of other stress-sensitive brain regions such as the prefrontal cortex should be considered.

The study has several limitations. The sample size was relatively modest, limiting the power to examine subgroup differences within specific diagnostic categories. The AH group had higher mean CPZ equivalents than the NAH group, although the findings held when controlling for this. The cross-sectional design prevents drawing causal inferences. The reliance on retrospective self-report for CSA assessment, while reliable in psychosis patients, could lead to underreporting. Finally, the study did not investigate amygdala subregions or functional connectivity.