Extending imaging-transcriptomics: from decoding group-based findings to individual brains
Poster No:
489
Submission Type:
Abstract Submission
Authors:
Min Tae Park1, Lena Palaniyappan1, Mallar Chakravarty1
Institutions:
1McGill University, Montreal, Canada
First Author:
Co-Author(s):
Introduction:
Imaging transcriptomics has been pivotal in linking variation in brain structure to spatial gene expression (in the Allen Human Brain Atlas (AHBA)1, for example), uncovering novel insights regarding the molecular mechanisms of neuroanaomical diversity2. These methods have focused on correlating group-based findings (i.e. effect sizes of case-control comparisons) with the AHBA. Building on this, a pertinent question arises: can these group-based correlations be replicated or observed in individual brains? Determining whether the generalizations at the group level with imaging transcriptomics also hold true at the individual level may provide a more nuanced and personalized understanding of brain function and neurobiology. We tested this hypothesis using brain MRI scans from healthy controls (HC) and patients with schizophrenia (SCZ).
Methods:
We hypothesized that we could infer gene expression, or transcriptomic "representation" maps using individual MRI data and multiple structural phenotypes. The goal is to identify transcriptomic signatures of brain structure in individual brains and identify broad pathway level correlations to brain structure.
Publicly available MRI datasets consisting of HC and SCZ were gathered, standard pipelines (CIVET) were used to measure cortical thickness (CT), surface area (SA), cortical volume (CV), and curvature (CC) across 40,892 vertices per hemisphere. After quality control, total sample size was n=1152 (HC=606, FEP=101, SCZ=445). We correlated individual phenotypes separately with AHBA gene expression measures across ~1,000 matched vertices for ~15k genes. We focused on CT-gene correlations given the recent focus. For each individual's gene correlations, we examined enrichment of gene sets including: 1) Schizophrenia genes (120 genes from 2022 PGC GWAS3), 2) Cell types, and 3) Biological pathways. Enrichment testing was done using the AUCell R package. Group differences in individual measures of enrichment were tested using mixed-effects models, taking age, sex as fixed effects and dataset as random effect.
Publicly available MRI datasets consisting of HC and SCZ were gathered, standard pipelines (CIVET) were used to measure cortical thickness (CT), surface area (SA), cortical volume (CV), and curvature (CC) across 40,892 vertices per hemisphere. After quality control, total sample size was n=1152 (HC=606, FEP=101, SCZ=445). We correlated individual phenotypes separately with AHBA gene expression measures across ~1,000 matched vertices for ~15k genes. We focused on CT-gene correlations given the recent focus. For each individual's gene correlations, we examined enrichment of gene sets including: 1) Schizophrenia genes (120 genes from 2022 PGC GWAS3), 2) Cell types, and 3) Biological pathways. Enrichment testing was done using the AUCell R package. Group differences in individual measures of enrichment were tested using mixed-effects models, taking age, sex as fixed effects and dataset as random effect.
Results:
Examining imaging-gene correlations across the four phenotypes reveals CT with the most variable correlations across genes (Figure). We did not find significant enrichment of SCZ genes in SCZ patients. Cell-type enrichment analysis demonstrates the oligodendrocyte precursor cell (OPC) enrichment as significantly greater in SCZ compared to HC (t=3.93, p=8.97E-05), replicating previous group-level comparison with imaging-transcriptomic analysis3. Astrocyte enrichment was significantly lesser in SCZ (t=-2.447, p=1.46E-02).
Conclusions:
Aligning individual MRIs to the AHBA may recapitulate imaging-transcriptomic correlations observed at the group level. Notably, we did not find significant enrichment of SCZ genes in SCZ patients compared to HC–suggesting a cumulative effect of group-based testing in gene enrichment analysis, while individual patient correlations with SCZ genes may be few in number. We replicated previous findings of cell-type enrichment in SCZ, specifically within OPC, astrocytes. Extending imaging transcriptomics into individuals appears to be feasible, and likely to provide more nuanced insights into individual neurobiology with possible clinical applications in the future.
Disorders of the Nervous System:
Psychiatric (eg. Depression, Anxiety, Schizophrenia) 1
Genetics:
Transcriptomics
Modeling and Analysis Methods:
Image Registration and Computational Anatomy 2
Multivariate Approaches
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
Cortical Anatomy and Brain Mapping
Keywords:
Cortex
Modeling
Morphometrics
MRI
Open Data
Schizophrenia
1|2Indicates the priority used for review
Provide references using author date format
2. Arnatkeviciute A, Fulcher BD, Bellgrove MA, Fornito A. Imaging Transcriptomics of Brain Disorders. Biological Psychiatry Global Open Science. 2021.
3. Trubetskoy V, Pardiñas AF, Qi T, et al. Mapping genomic loci implicates genes and synaptic biology in schizophrenia. Nature. 2022;604(7906):502-508.
4. Di Biase MA, Geaghan MP, Reay WR, et al. Cell type-specific manifestations of cortical thickness heterogeneity in schizophrenia. Molecular psychiatry. 2022;27(4):2052-2060.