White matter microstructure alterations in early psychosis and schizophrenia

Abstract

Studies on schizophrenia feature diffusion magnetic resonance imaging (dMRI) to investigate white matter (WM) anomalies. The heterogeneity in the possible interpretations of typical Diffusion Tensor Imaging (DTI) metrics highlights the importance of increasing their specificity. Here, we characterize WM pathology in early psychosis (EP) and schizophrenia (SZ) with increased specificity using advanced dMRI: Diffusion Kurtosis Imaging and the biophysical model White Matter Tract Integrity - Watson (WMTI-W). This enables us to better characterize WM abnormalities, while preserving good sensitivity to group differences, and relate them to the current literature (ENIGMA-schizophrenia), patient's clinical characteristics and symptomatology. dMRI-derived microstructure features were extracted from all of WM and from individual regions of interest in 275 individuals. 93 subjects diagnosed with EP and 47 with SZ were compared respectively to 135 age-range matched healthy controls (HC). WM DTI diffusivities were higher, while kurtosis was lower in EP vs HC and in SZ vs HC. Differences were more widespread in EP than SZ. The regional alterations found in our cohort matched the DTI patterns found in ENIGMA-schizophrenia. WMTI-W model parameters indicate that the WM alterations in patients come primarily from the extra-axonal compartment, consistent with abnormal myelin integrity in the disease pathology. The direct link between WM alterations and symptomatology is, however, limited.

Fig. 1. Strip plot of the group comparisons.

Each clinical group is compared to its respective HC group. A–D: DTI metrics, E–F: DKI metric, H–L: WMTI-W metrics. *:p ≤ 5e-2, **:p ≤ 1e-2, ***:p ≤ 1e-3, ****:p ≤ 1e-4.

Fig. 2. Heatmaps and brainplots of the JHU ROI group comparison.

Top: group comparisons (y-axis right) heatmap of the effect size, thresholded for significance, for each dMRI metric (y-axis left) and each region of interest (x-axis). Bottom: brainplots of the same comparisons. The colorbar is shared. Red: clinical group > HC, Blue: clinical group < HC. NS: p > 0.05, *:p ≤ 5e-2, **:p ≤ 1e-2, ***:p ≤ 1e-3, ****:p ≤ 1e-4. Numeric p-values and effect sizes can be found in Table S5, S6. Abbreviations: L: left; R: right; MCP: Middle cerebellar peduncle; PCT: Pontine crossing tract; GCC/BCC/SCC: Genu/Body/Splenium of corpus callosum; ; FX: Fornix (column and body); CST: Corticospinal tract; ML: Medial lemniscus; ICP/SCP: Inferior/Superior cerebellar peduncle; CP: Cerebral peduncle; ALIC/PLIC/RPIC: Anterior limb/Posterior limb/ Retrolenticular part of internal capsule; ACR/SCR/PCR: Anterior/Superior/Posterior corona radiata; PTR: Posterior thalamic radiation; SS: Sagittal stratum; EC: External capsule; CGC: Cingulum (cingulate gyrus); CGH: Cingulum (hippocampus); FX/ST: Fornix / Stria terminalis; SLF: Superior longitudinal fasciculus; SFO/IFO: Superior/Inferior fronto-occipital fasciculus; UNC: Uncinate fasciculus; TAP: Tapetum.

Fig. 3. ENIGMA versus Lausanne psychosis (LSP) regional effect sizes (Cohen’s d).

Plots are divided by EP (top row, A–E) and SZ (bottom row, F–K). Here, “averageWM” refers to ENIGMA average WM skeleton while for LSP it corresponds to the WM core (average of all JHU ROIs combined in one mask).

Fig. 4. Average effect sizes across regions of interest showing significant changes in the EP and SZ groups.

Average effect sizes of significant ROIs for each dMRI metric in EP vs HC_Young (A) and SZ vs HC_Old (B). The black horizontal lines represent the confidence intervals.