Connectomic intermediate phenotypes for psychiatric disorders

Abstract

Psychiatric disorders are phenotypically heterogeneous entities with a complex genetic basis. To mitigate this complexity, many investigators study so-called intermediate phenotypes (IPs) that putatively provide a more direct index of the physiological effects of candidate genetic risk variants than overt psychiatric syndromes. Magnetic resonance imaging (MRI) is a particularly popular technique for measuring such phenotypes because it allows interrogation of diverse aspects of brain structure and function in vivo. Much of this work however, has focused on relatively simple measures that quantify variations in the physiology or tissue integrity of specific brain regions in isolation, contradicting an emerging consensus that most major psychiatric disorders do not arise from isolated dysfunction in one or a few brain regions, but rather from disturbed interactions within and between distributed neural circuits; i.e., they are disorders of brain connectivity. The recent proliferation of new MRI techniques for comprehensively mapping the entire connectivity architecture of the brain, termed the human connectome, has provided a rich repertoire of tools for understanding how genetic variants implicated in mental disorder impact distinct neural circuits. In this article, we review research using these connectomic techniques to understand how genetic variation influences the connectivity and topology of human brain networks. We highlight recent evidence from twin and imaging genetics studies suggesting that the penetrance of candidate risk variants for mental illness, such as those in SLC6A4, MAOA, ZNF804A, and APOE, may be higher for IPs characterized at the level of distributed neural systems than at the level of spatially localized brain regions. The findings indicate that imaging connectomics provides a powerful framework for understanding how genetic risk for psychiatric disease is expressed through altered structure and function of the human connectome.

Keywords: Alzheimer’s disease; anxiety; complex; default mode; depression; endophenotype; graph analysis; schizophrenia.

Figure 1

A schematic overview of the different neuroimaging techniques available for measuring structural and functional properties of the human connectome. DCM, dynamic causal modeling; DWI, diffusion-weighted imaging; GCA, Granger causality analysis; ICA, independent component analysis; MCA, morphometric covariance analysis; PPI, psychophysiological interactions; PWC, pair-wise correlations; TBSS, tract-based spatial statistics; T1, T1-weighted imaging; VBA, voxel-based analysis; WM, white matter.

Figure 2

Genetic influences on functional connectivity of fronto-limbic circuits. Cortical surface renderings display the stereotactic peaks of regions in prefrontal cortex where functional connectivity with the amygdala is impacted by common genetic variants. Different colors denote the different genes studied in relation to fronto-limbic functional connectivity; namely, SLC6A4 (Heinz et al., ; Pezawas et al., ; Schardt et al., 2010), MAOA (Buckholtz et al., 2008), COMT (Rasch et al., 2010), and DRD2 (Blasi et al., 2009). Arrows indicate whether functional connectivity was increased (↑) or decreased (↓) in carriers of the putative risk allele for each gene. The distinction between cyan and blue foci for SLC6A4 studies differentiates a single study examining functional connectivity during cognitive regulation of emotion (Schardt et al., 2010) from others involving passive perception of emotional stimuli. This distinction illustrates how the risk allele of 5-HTTPLR polymorphism of this gene can be associated with either increased or decreased functional connectivity with the amygdala in adjacent regions of right pre-genual ACC, depending on task context.

Figure 3

Illustration of the hippocampal region where functional connectivity with a seed in right dorsolateral PFC was influenced by ZNF804A variation and was found to differ between schizophrenia patients, their unaffected relatives and healthy controls. The two bar charts plot parameter estimates for fronto-hippocampal functional connectivity as a function of ZNF804A in healthy controls (middle), and in controls, patients, and siblings (right). Carriers of the A risk allele showed an increasing trend toward greater negative functional connectivity, as did patients with schizophrenia. Image adapted from Rasetti et al. (2011).

Figure 4

Cortical regions showing statistically significant (p < 0.05, corrected) heritability for regional cost-efficiency of network functional connectivity. Arrows highlight regions showing significant genetic effects in both cerebral hemispheres. DMPFC, dorso-medial prefrontal cortex; PCC, posterior cingulate cortex; SFG, superior frontal gyrus; pMFG, posterior middle frontal gyrus; PCG, post-central gyrus; SPL, superior parietal lobule; pSTS, posterior superior temporal sulcus. Left hemisphere is presented on the right. Image reproduced from Fornito et al. (2011b).