Identifying and validating subtypes within major psychiatric disorders based on frontal-posterior functional imbalance via deep learning

Abstract

Converging evidence increasingly implicates shared etiologic and pathophysiological characteristics among major psychiatric disorders (MPDs), such as schizophrenia (SZ), bipolar disorder (BD), and major depressive disorder (MDD). Examining the neurobiology of the psychotic-affective spectrum may greatly advance biological determination of psychiatric diagnosis, which is critical for the development of more effective treatments. In this study, ensemble clustering was developed to identify subtypes within a trans-diagnostic sample of MPDs. Whole brain amplitude of low-frequency fluctuations (ALFF) was used to extract the low-dimensional features for clustering in a total of 944 participants: 581 psychiatric patients (193 with SZ, 171 with BD, and 217 with MDD) and 363 healthy controls (HC). We identified two subtypes with differentiating patterns of functional imbalance between frontal and posterior brain regions, as compared to HC: (1) Archetypal MPDs (60% of MPDs) had increased frontal and decreased posterior ALFF, and decreased cortical thickness and white matter integrity in multiple brain regions that were associated with increased polygenic risk scores and enriched risk gene expression in brain tissues; (2) Atypical MPDs (40% of MPDs) had decreased frontal and increased posterior ALFF with no associated alterations in validity measures. Medicated Archetypal MPDs had lower symptom severity than their unmedicated counterparts; whereas medicated and unmedicated Atypical MPDs had no differences in symptom scores. Our findings suggest that frontal versus posterior functional imbalance as measured by ALFF is a novel putative trans-diagnostic biomarker differentiating subtypes of MPDs that could have implications for precision medicine.

Conflict of interest statement

The authors declare that they have no conflict of interest.

© 2020. The Author(s).

Figures

Fig. 1. Schematic of using deep learning-based hierarchical clustering to define clusters of MPDs.

Step one: identification of significant functional alterations in MPDs and using AutoEncoder to reduce the dimension of the identified alterations to d ∈ [2,10]. Step two: for each of the nine low-dimensional data from step one, we obtained nine different class labels based on clustering analyses, and five clusters (cluster A, B, C, D, and E) were identified. Step three: we performed the clusters merging process according to six runs of clustering and obtained two final subtypes. Furthermore, the subtypes varied in patterns of amplitude of low-frequency fluctuation alterations as compared to HC (voxel p < 0.001 with Gaussian random field correction for cluster p < 0.05). MPD major psychiatric disorder; HC healthy control; L left; R right; d dimension.

Fig. 2. Significant differences in (a) cortical thickness and (b) white matter integrity between Archetypal MPDs and healthy controls.

Significance level was set to voxel p < 0.001 with Gaussian random field correction for cluster p < 0.05. The color bar represents t value. MPD major psychiatric disorder.

Fig. 3. The variance (y-axis) of case-control status explained by the PRS-SZBD and PRS-MDD in Archetypal and Atypical MPDs.

x-axis represents p value threshold, y-axis represents PRS model fit: R2 (Nagelkerke’s). The bars represent ten best-fit PRS scores calculated at different p value threshold. ***p < 0.001; **p < 0.01. PRS-SZBD, polygenetic risk score of schizophrenia and bipolar disorder, PRS-MDD, polygenetic risk score of major depressive disorder. MPD major psychiatric disorder.

Fig. 4. Differentially expressed risk genes across 53 tissues in (a) Archetypal and (b) Atypical MPDs.

MPD major psychiatric disorder.

Fig. 5. Significant differences in HAMD factors and BPRS factor scores between medicated and unmedicated patients in Archetypal and Atypical MPDs.

The significance level was set to p < 0.05 with false discovery rate correction. Vertical black lines show the standard errors of the means. ***p < 0.001; **p < 0.01. HAMD Hamilton Depression Scale; BPRS Brief Psychiatric Rating Scale; MPD major psychiatric disorder.