fMRI fluctuations within the language network are correlated with severity of hallucinatory symptoms in schizophrenia

Schizophrenia (SZ) is a complex psychiatric disorder with heterogeneous symptoms and unclear neurobiological mechanisms. Auditory verbal hallucinations (AVH), a distressing symptom affecting 60-90% of SZ patients, are a key focus of this study. Crow's hypothesis posits that AVH are linked to decreased left hemispheric dominance for language. Research suggests alterations in hemispheric communication, especially within the language network (LN), in SZ patients. Resting-state fMRI studies have investigated the relationship between brain function and AVH vulnerability. The resting state provides a baseline measure of spontaneous brain activity organized into networks, including the default mode network (DMN), which overlaps with networks modulated during tasks, such as the LN. The current study aimed to investigate the neural correlates of AVH severity in the LN of SZ patients, comparing two subgroups based on AVH vulnerability (AVH/D+ and AVH/D−) with healthy controls (HC). The study used a novel approach combining fMRI functional connectivity and spectral analysis (fALFF) to assess both network connectivity and the spectral properties of the signal, which may provide additional information beyond traditional functional connectivity analyses. This study aimed to identify whether AVH/D+ patients show aberrant LN connectivity, potentially characterized by right hemisphere recruitment, while HC and AVH/D− groups would show a typical left-lateralized LN pattern. The fALFF analysis was expected to reveal correlations between specific frequency bands and the severity of hallucinatory (P3) and delusional (P1) symptoms measured using the Positive and Negative Syndrome Scale (PANSS).

Several studies have linked auditory verbal hallucinations (AVH) in schizophrenia to decreased functional left hemispheric dominance, particularly in speech processing areas of the left temporal lobe. Research has also pointed towards alterations in hemispheric brain communication, especially within the language network (LN), in schizophrenia patients across various tasks and linguistic experimental approaches. A growing body of research has directly linked resting-state brain function to AVH vulnerability. The resting-state condition, reflecting spontaneous functional organization into networks such as the default mode network (DMN), offers a unique perspective on brain activity not influenced by external demands. The DMN, while primarily a resting-state network, overlaps with networks modulated during active tasks, including the LN. The prevailing view considers schizophrenia as a disconnection syndrome, with Crow's hypothesis emphasizing the central role of language and its underlying network in this disconnection. The complex integration and connection of cortical and subcortical areas required for language processing make the LN a prime target for investigation in schizophrenia.

This study used de-identified data from two subgroups of Norwegian schizophrenia patients: AVH/D+ (predominantly positive symptoms, PANSS P3 score ≥ 4) and AVH/D− (predominantly negative symptoms, PANSS P3 score ≤ 4). A further criterion involving PANSS P1 (delusions) score was used to resolve cases with a P3 score of 4. Seventeen patients per subgroup were included. An age-matched healthy control (HC) group (n=17) was also included. All patients were on second-generation antipsychotics. MRI data (resting-state fMRI and T1-weighted structural images) were acquired on a 3T GE Discovery 750 scanner. fMRI data preprocessing involved spatial alignment, motion correction, bias field correction, spatial smoothing, and co-registration to standard space. Functional connectivity analysis used Independent Component Analysis (ICA) to extract the language network (LN) spatial pattern and time course at the single-subject level. Between-group differences in LN spatial maps were assessed using one-sample and two-sample t-tests, with Benjamini-Hochberg false discovery rate (BH-FDR) correction for multiple comparisons. Fractional amplitude of low-frequency fluctuations (fALFF) analysis was performed on the LN time series, divided into four frequency bands (slow-5, slow-4, slow-3, slow-2). fALFF values were normalized with respect to canonical fALFF. Spearman correlation analyses investigated the relationship between normalized fALFF values and PANSS P1 (delusions) and P3 (hallucinations) subscales. Post hoc analyses examined fALFF in regions of interest (ROIs) in the right and left frontal operculum, using principal component analysis (PCA) to derive a representative signal from each ROI. Finally, a voxel-based morphometry (VBM) analysis was conducted on structural MRI data to investigate potential underlying anatomical differences that might confound the functional findings.

No significant socio-demographic differences were found between the HC and SZ groups, although AVH/D− patients had significantly higher dosages of second-generation antipsychotics than AVH/D+ patients. PANSS scores (P1 and P3) were significantly higher in the AVH/D+ group. ICA revealed that the LN in AVH/D+ patients included atypical extra regions in the right hemisphere (homologous of Broca's area/frontal operculum and anterior insula), unlike the HC and AVH/D− groups which showed typical left-lateralized LN. Between-group comparisons showed significantly greater connectivity in the right opercular part of the inferior frontal gyrus (BA 47) and decreased connectivity in the left triangular part of the inferior frontal gyrus (BA 45) in AVH/D+ patients compared to HC. These differences were also observed between AVH/D+ and AVH/D− groups. The fALFF analysis showed that AVH/D+ patients had significantly lower slow-frequency fluctuation amplitudes compared to HC in the 0.012–0.018 Hz range, and significantly higher amplitudes in the 0.03 and 0.036 Hz range. AVH/D− patients showed significant differences with HC in the 0.06–0.08 Hz range. The AVH/D+ and AVH/D− groups differed significantly at around 0.06 Hz. Correlation analyses revealed that for AVH/D− patients, higher amplitudes of LN fALFF (Slow-4) correlated with greater delusion severity (P1). Post-hoc ROI analysis on AVH/D+ patients found significant negative correlations between P1 scores and fALFF Slow-4 and positive correlations between P3 scores and fALFF Slow-5 in the right frontal operculum. Finally, no significant structural differences were found between groups in the ROIs examined, indicating that the observed functional differences were unlikely due to underlying anatomical aberrations.

The findings support the hypothesis that increased vulnerability to hallucination in schizophrenia is associated with alterations in the language network (LN). The atypical right hemisphere activation in AVH/D+ patients, particularly involving the insula and Broca's homologous area, is consistent with Crow's hypothesis of reduced left hemisphere dominance in hallucinatory psychosis. The insula's role in integrating sensory input with internal emotional processing suggests its involvement in self-other distinction, making its dysfunction a potential contributor to AVH. The fALFF analysis revealed distinct patterns of correlation between specific frequency bands and PANSS scores. While AVH/D− patients showed a relationship between increased left-hemisphere LN fALFF and delusion severity, AVH/D+ patients showed an inverted relationship in the right frontal operculum, suggesting that spectral power alterations in this area may be linked to hallucination severity. The lack of significant structural differences emphasizes the importance of functional alterations in explaining vulnerability to AVH. The right frontal operculum/insula's role as a critical hub connecting various brain areas involved in complex cognitive processes strengthens the importance of these findings.

This study provides evidence that alterations in the language network, particularly atypical right hemisphere activation and specific spectral power fluctuations within this network, are associated with the severity of hallucinatory and delusional symptoms in schizophrenia. The insula's key role in self-other distinction and its involvement in the observed patterns highlight its potential importance in the pathophysiology of AVH. Future research should investigate AVH characteristics with more granularity, explore the hallucinatory phenomenon during acute episodes, and use larger datasets to corroborate these findings. Further investigation into the specific frequency bands and their relationship to different symptom dimensions warrants further investigation.

The study used a relatively small sample size, which might limit the generalizability of the findings. The cross-sectional nature of the study prevents causal inferences. The reliance on resting-state fMRI may not fully capture the dynamic nature of brain activity during actual hallucinatory experiences. The fact that all patients were on medication could potentially confound the results, although the study tried to mitigate this through statistical controls and the use of a comparison group of similar patients. Finally, the emphasis on auditory verbal hallucinations (AVH) might limit the generalizability to other types of hallucinations present in schizophrenia.