Cell type-specific manifestations of cortical thickness heterogeneity in schizophrenia

Abstract

Brain morphology differs markedly between individuals with schizophrenia, but the cellular and genetic basis of this heterogeneity is poorly understood. Here, we sought to determine whether cortical thickness (CTh) heterogeneity in schizophrenia relates to interregional variation in distinct neural cell types, as inferred from established gene expression data and person-specific genomic variation. This study comprised 1849 participants in total, including a discovery (140 cases and 1267 controls) and a validation cohort (335 cases and 185 controls). To characterize CTh heterogeneity, normative ranges were established for 34 cortical regions and the extent of deviation from these ranges was measured for each individual with schizophrenia. CTh deviations were explained by interregional gene expression levels of five out of seven neural cell types examined: (1) astrocytes; (2) endothelial cells; (3) oligodendrocyte progenitor cells (OPCs); (4) excitatory neurons; and (5) inhibitory neurons. Regional alignment between CTh alterations with cell type transcriptional maps distinguished broad patient subtypes, which were validated against genomic data drawn from the same individuals. In a predominantly neuronal/endothelial subtype (22% of patients), CTh deviations covaried with polygenic risk for schizophrenia (sczPRS) calculated specifically from genes marking neuronal and endothelial cells (r = -0.40, p = 0.010). Whereas, in a predominantly glia/OPC subtype (43% of patients), CTh deviations covaried with sczPRS calculated from glia and OPC-linked genes (r = -0.30, p = 0.028). This multi-scale analysis of genomic, transcriptomic, and brain phenotypic data may indicate that CTh heterogeneity in schizophrenia relates to inter-individual variation in cell-type specific functions. Decomposing heterogeneity in relation to cortical cell types enables prioritization of schizophrenia subsets for future disease modeling efforts.

Fig. 1. Schematic diagram of methodology.

a Overview of the methodology and datasets. Note that steps b to d were completed for two independent datasets: Human Connectome Project (HCP; discovery) and Australian Schizophrenia Research Bank (ASRB; validation). b Quantile regression was used to establish normative ranges of individual CTh variation, given age and sex. Percentile curves (5%, 50%, 95%) are shown as a function of age for CTh in an example cortical region. Shading denotes 95% confidence intervals. For each cortical region, individuals with schizophrenia were compared to the normative range established for their age and sex. The vector shown provides a putative CTh deviation profile that summarizes individual differences across 34 brain regions. c Cell-class gene-expression brain maps reflected the mean expression of seven cell-type gene sets (columns) in each of the 34 (left hemisphere) cortical regions (rows). d For each patient (rows), a correlation coefficient (matrix elements) quantified the extent of spatial coupling between interregional cortical deviation scores and the 7 cell-class gene expression maps (columns). The Ansari-Bradley test was used to evaluate whether the observed distribution of correlations significantly differed from a null distribution. e Ward’s linkage was used to cluster individuals into broad cell-based subtypes based on spatial coupling between regional CTh deviations and regional transcriptomic maps. f Partial correlations evaluated whether polygenic cell scores (e.g., neurons) covaried with the severity of cortical thickness deviations among individuals comprising the broad cell-based subtypes.

Fig. 2. Cell type-specific gene expression maps and mean CTh variation in healthy controls.

a Cortical renderings display mean Cortical thickness (CTh) for each region across all healthy individuals, respectively for HCP (discovery) and ASRB (validation) cohorts. b Cortical renderings display standardized (z-scores) for cell type-specific gene expression maps. c Scatter plots display associations between standardized cell type-specific gene expression maps and regional mean CTh in healthy controls. Each datapoint (34 in total) reflects a left hemispheric brain region from the Deskian-Killiany atlas, where mean CTh was computed across all healthy controls in the discovery or validation cohorts.

Fig. 3. Deviation from normative ranges of CTh variation in individuals with schizophrenia.

a Cortical renderings display the mean deviation score for each region across all individuals with schizophrenia, respectively for HCP and ASRB datasets. b Percentile curves for example regions. The 5th (red curve), 50th (black) and 95th (blue) percentiles quantify the range of variation among healthy individuals (dark dots) in the cortical thickness (CTh), as a function of age (horizontal axis) and sex. For each cortical region, individuals with schizophrenia were positioned on the normative percentile charts and then categorized as either: (i) normal (black); (ii) supra-normal (blue cross); or, (iii) infra-normal (red cross). The measurement unit of CTh is millimeters and age is quantified in years. Shading indicates 95% confidence intervals, estimated with bootstrapping (n = 1000). c Loci of variation in a 31-year old male (upper panel) and 65-year old female (lower panel) with schizophrenia.

Fig. 4. Association between cell type-specific gene expression patterns and CTh deviations in schizophrenia.

a Density functions display the distribution of gene expression—CTh correlation coefficients (ranging from −1 to 1) for genes marking specific cell types. Plots display the null distribution of correlation coefficients (black), and the observed correlation coefficients (red) in patients comprising the HCP (discovery) and ASRB (validation) cohorts. b Plots display observed (red) and null cumulative distribution functions (blue) of correlation coefficients between regional gene expression and regional deviations in CTh. Results are presented for HCP (discovery) and ASRB (validation) cohorts. Alpha denotes a significant difference after FDR correction. These findings illustrate that the spatial patterning of CTh deviations significantly relate to expression gradients of genes that regulate specific cell classes. For example, the first plot shows that regional gene expression across genes marking astrocytes correlate more strongly with CTh deviation profiles in patients, relative to random deviation profiles (i.e., after permuting regional deviation values).

Fig. 5. Cell-based patient subtypes and genomic validation.

Results are shown for the validation (ASRB) cohort a Person-specific correlations between cortical deviations with interregional cell type-specific gene expression maps were clustered into three cell-based subtypes. b Boxplots show characteristic gene expression-CTh deviation association patterns for each subtype. Colored boxes denote negative associations (i.e., where CTh loss maps onto higher cell type gene expression) that significantly differ from zero (pFDR < 0.05). Box edges indicate 25th and 75th percentiles of inter-individual variation in standardized gene expression for each cell type. Central mark indicates median, whisker extend to the most extreme datapoints, and circles denote outliers. c The top line plot demonstrates that the maximum gap criterion occurs at three clusters, which is more than one standard error from the next maximum gap value. The bottom plot shows the t-Distributed Stochastic Neighbor Embedding (t-SNE), whereby person-specific points are embedded into three clusters in a way that respects similarities between points. The pie chart displays the portion of schizophrenia subjects comprising each subtype. d Brain renderings display mean CTh deviation profiles across individuals from each subtype, respectively. The boxplot displays mean cortical thickness deviations across the three subtypes (bars = median, boxes = lower and upper quartiles and circles = outliers computed from the interquartile range). e Scatterplots (left column) present correlations between polygenic scores (y-axes) and cortical thickness deviations (x-axes), colored according to cell-based subtype (e.g., dark blue datapoints represent individuals comprising Subtype I-Neuronal). Shaded areas represent the 95% confidence interval, colored according to subtype. Boxplots (right column) display r values for correlations evaluated using schizophrenia subjects sorted by subtype membership (ranging from zero to one). For example, ‘Soft Subtype I’ shows that correlation strength increases as a function of membership to Subtype I—i.e., the r value decreases as more subjects with low membership scores are excluded from the correlation analyses.