Unraveling the Link between CNVs, General Cognition, and Individual Neuroimaging Deviaton Scores.

865 

Abstract Submission 

charlotte fraza¹, Andre Marquand², Rune Boen³, Ida Sønderby³

¹Donders Institute for Brain, Cognition and Behaviour, Nijmegen, gelderland, ²Radboud University Nijmegen, Nijmegen, Gelderland, ³Oslo University Hospital, Oslo, Norway

charlotte fraza  
Donders Institute for Brain, Cognition and Behaviour
Nijmegen, gelderland

Andre Marquand  
Radboud University Nijmegen
Nijmegen, Gelderland

Rune Boen  
Oslo University Hospital
Oslo, Norway

Ida Sønderby  
Oslo University Hospital
Oslo, Norway

Copy number variations (CNVs) are genetic variants that involve a deletion or a duplication of a larger segment of the genome. Certain CNVs yield a higher risk of developing neurodevelopmental and psychiatric disorders, but with considerable differences in outcome between carriers (Malhotra & Sebat, 2012; Modenato et al., 2021; Sanders et al., 2019). Understanding their individual-level effects on the brain and behavior is pivotal for unraveling their intricate contributions to cognition and mental health. While conventional studies often contrast cases from controls, this is limited to groups that have sufficient sample sizes and thus often excludes mapping of the rare, pathogenic CNVs with high impact. Normative modeling offers a solution by quantifying individual deviations from reference populations which may aid in decoding complex brain disorders (Marquand et al., 2016, 2019).
This study employs normative modeling to explore the personalized impacts of pathogenic CNVs on brain function and behavior, highlighting the 1q21.1 distal deletion and duplication. By linking brain deviation scores from a reference model to cognition and investigating shared alterations among individuals with similar CNVs, this research gives insights into the nuanced role of CNVs in shaping personalized brain deviation maps.

An overview of our analytic workflow is presented in Fig. 1. In brief, we fitted normative models for several image derived phenotypes (IDPs), derived from three primary imaging modalities: structural, functional, and diffusion MR data, and the Jacobian determinants to create a comprehensive exploration of the effect of CNVs on the brain. We selected specific CNVs that have previously been proposed to be pathogenic and that have been associated with a decrease in cognition (Kendall et al., 2017, 2019). Fig. 1A visually outlines our study workflow: i. We mapped participants with pathogenic CNVs, such that they can be placed in our test set. ii. We created the normative models for each individual IDP taking into account several covariates, such as age and sex. iii. We counted the number of extreme deviations for each participant and map where participants with a pathogenic CNV lie on this distribution. iv. We correlated the extreme z-scores with cognition.

We delved into individual brain deviation scores associated with specific CNVs, with a special emphasis on 1q21.1 distal deletion and duplication, see figure 2. We counted extreme deviations (|Z|>2) and provided maps of deviation scores that highlighted regions with pronounced alterations. Our analysis revealed that individuals with 1q21.1del displayed more positive deviations indicating more volume expansions, while those with 1q21.1dup exhibited more negative deviations indicating more volume contractions. We conducted a joint analysis to uncover common and divergent deviation scores in specific brain regions, and this approach proved applicable to various other pathogenic CNVs.
Furthermore, we explored the relationship between brain deviation scores and cognitive deficits. Our hypothesis posited that certain CNVs might impact cognition, especially in the absence of protective factors. Our Pearson correlation analysis indicated that a higher number of extreme negative deviations was associated with a lower general cognitive ability, while no significant correlation was observed for extreme positive deviations.

Our study emphasizes the importance of individual-level analysis over traditional case-control thinking within pathogenic CNV research and neuroimaging. We advocate for an "individual patient first" approach in psychiatry, acknowledging the diversity and comorbidities in mental disorders and their underlying causal pathways. We believe comprehensive normative models can capture the inherent heterogeneity present in participants with pathogenic CNVs, and the effects of these CNVs on brain morphology, providing a valuable perspective for future research.

Genetic Modeling and Analysis Methods ¹

Lifespan Development Other ²

Bayesian Modeling

Machine Learning

Modeling

MRI