Association of GRM3 polymorphism with white matter integrity in schizophrenia

Abstract

Background: While the functional disconnectivity hypothesis of schizophrenia has received considerable attention, fewer studies have investigated the contribution of genotype to structural connectivity between brain regions either in schizophrenia patients or in healthy controls. In this study, we obtained diffusion tensor imaging (DTI) data and genome-wide single nucleotide polymorphism (SNP) data from 74 cases and 87 age- and gender-matched controls.

Methods: We used independent component analysis (ICA) to analyze fractional anisotropy (FA) values and correlated FA values with 121 SNPs in genes associated with myelination and/or schizophrenia risk.

Results: Using ICA, we identified 6 maximally independent components in which the majority of the voxels corresponded to known white matter (WM) tracts. Among these WM-enriched components, two had FA values that were significantly decreased in patients. In addition, we examined the relationship between FA values and genotype and found that a SNP located in the intronic region of the metabotropic glutamate receptor 3 gene, GRM3, shows a significant correlation with FA values in a component containing tracts from the cortico-cerebellar-thalamic-cortical circuit of patients but not controls.

Conclusions: Our findings strengthen the evidence for an association between GRM3 genotype and schizophrenia and suggest a role for glutamate neurotransmission in the establishment and maintenance of myelinated fibers.

Keywords: Cortico–cerebellar–thalamic–cortical circuit; DTI; GRM3; Independent component analysis; Schizophrenia; mGluR3.

Figure 1. Threshold map for WM-enriched components identified by ICA on multi-site DTI data

(A–F). Color scale represents the Z value for each voxel’s ICA score. The higher scoring voxels from each biologically relevant component tend to be distributed symmetrically and along regions of white matter tracts.

Figure 2. Correlation of ICA loading coefficients with lifetime CPZ dose-year equivalent use

ICA loading coefficients in components B (panel A) and E (panel B) showed a significant association with lifetime CPZ use in patients when including site as a covariate (component B corrected p=0.0131; component E corrected p =0.00599).

Figure 3. Effect of rs7808623 genotype on component C ICA loading coefficients

ICA loading coefficients for each subject in the patient group plotted as a function of rs7808623 genotype and corrected p with site as a covariate (p=0.0252).

Figure 4. Threshold maps of component C

Coronal slices in fixed intervals showing threshold map starting anteriorly in the top right (A). Sagittal section without threshold map overlay showing interior and posterior boundaries of coronal sections (B). Color scale represents the Z value for each voxel’s ICA score.