Neurobiology of bipolar subtypes: multivariate structural neuroimaging approaches

Debates persist regarding the optimal approach to manage the heterogeneity behind the broad concept of bipolar disorder (BD), with the precise neurobiological mechanisms differentiating the main BD subtypes, BD type I (BD-I) and BD type II (BD-II), remain a subject of ongoing research. The exploration of structural brain networks, coupled with an in-depth examination of the interplay between network covariance patterns, clinical, psychological functioning, and affective temperaments, has not been systematically investigated across the BD subtypes. Thirty BD-I, 30 BD-II, and 45 age- and sex-matched healthy controls (HC) underwent an MRI scan. For the neuropsychological measures, the Tower of Hanoi to investigate executive functions, the Facial Emotion Recognition (FER) task for emotional processing, and the Temperament Evaluation of Memphis, Pisa, Paris, and San Diego Scale to investigate affective temperaments were administered to participants. Clinical assessment was conducted, and a detailed history of mood disorders and pharmacological treatments were recorded. Gray matter volume (GMV) and white matter were estimated using voxel-based morphometry and fractional anisotropy (FA) analysis, respectively. Source-based morphometry (SBM) was performed to extract structural covariation patterns of GMV and white matter FA. GMV, FA values, and SBM components associated with morphometric and FA differences were compared across BD-I, BD-II, and HC. GMV of the cortico-limbic regions implicated in emotional processing was estimated and their relationship with FER performance was investigated using network analysis. To explore the relationships between structural changes, clinical data, executive and emotional processing measures, Pearson's or Spearman's correlation were performed. BD-I exhibited reduced GMV in the temporo-insular-parieto-occipital cortex and in the cerebellum compared with HC, while no significant differences between BD-II and HC nor between BD subtypes emerged. An SBM component encompassing the prefrontal-temporal-occipital network displayed significantly lower GMV in BD-I compared to HC, but not between the other groups. The reduced structural network covariance in BD-I was associated with the number of previous manic episodes and with worse executive performance. Compared with BD-II, BD-I showed a general impairment in FER performance, which was correlated with temporal-occipital GMV loss. BD-I displayed the poorest sadness-related FER performance relative to the other groups. In BD-I, sadness-related FER performance was negatively correlated with illness duration and the number of previous manic episodes and positively associated with global functioning. Overall structure of the network of BD subtypes was altered compared with HC, with BD-I showing a reduced GMV interrelationship in the frontal–insular–occipital regions and greater edge strength between sadness-related FER performance and amygdala GMV relative to HC. Compared with HC, BD-I and BD-II showed significant FA reduction encompassing the corpus callosum, the genu of the corpus callosum, the cingulum and parahippocampal circumvolution bilaterally, and parieto-occipital regions bilaterally. BD-II had two clusters of significantly greater FA located in the genu of the corpus callosum and in the right parahippocampal circumvolution relative to BD-I. In BD-I, the SBM component of lower cingolo-callosal-parahippocampal FA covariance (IC1 loadings) had a significant positive correlation with the irritable temperament score, which was significantly higher in BD-I than HC. Using multivariate approaches, incorporating clinical, behavioral, and neuroimaging data analysis, we observed several structural and behavioral differences between BD-I and BD-II that confirm the clinical and nosological differentiation between the main BD subtypes. The identified networks can be targeted for more accurate prognosis prediction and tailored interventions.