Email Icon Facebook icon Twitter Icon GooglePlus Icon Contact

User Top Menu

SfN 2011—Connectomics and Schizophrenia (Part 1)

2 Apr 2012

Editor's Note: As the final installment in our coverage of the 2011 Society for Neuroscience meeting, held 12-16 November 2011 in Washington, DC, we bring you an illuminating two-part survey of connectomics studies, written by volunteer meeting reporter Scott Bolkan of Columbia University. See also Part 2.

3 April 2012. On display at last year’s Society for Neuroscience meeting in Washington, DC, was a burgeoning interest in the field of connectomics. Analogous to the Human Genome Project, which determined set out to chart every one of the three billion chemical base pairs that comprisein human DNAchromosomes, connectomics aims to physically map the complete set of neural circuits that collect, process, and archive information in the brain. This growing field, and the techniques rapidly developing alongside it, has the potential for testing a long-standing hypothesis on the pathophysiology of schizophrenia —the "disconnection" hypothesis.

The disconnection hypothesis of schizophrenia has its origins in observations of postmortem anatomy and clinical intuition. While the hypothesis as originally stated implies reduced anatomical connectivity, this idea has recently been generalized to the concept of "disconnectivity." Disconnectivity can involve an abnormal relationship between neurons in multiple scales of space and time—from the disruption of local microcircuits to the disruption of brainwide neural communication (reviewed by Stephan et al., 2009).

Connectomics is a field that can be distinguished by the scale of analysis or the model organism used. At the nano- or micro-scale, single neurons and the complete set of synapses they make can be determined with techniques such as serial block free scanning electron microscopy. While suitable for model organisms like the fly (104 neurons) and, arguably, the mouse (108 neurons), the invasiveness of the technique and the complexity of the human brain (1011 neurons and 1015 synapses) make synapse-resolution mapping of human brains unfeasible.

Efforts to create a human connectome have thus employed a variety of non-invasive imaging techniques capable of detailing a macro-scale map of anatomically distinct brain regions and the interregional pathways connecting them (see the original human connectome proposal of Sporns et al., 2005). Of particular importance is the technique of diffusion-weighted magnetic resonance imaging (MRI), such as diffusion tensor imaging (DTI). With DTI, researchers obtain voxelwise measures of the diffusion of water molecules, and extract a value of density probability and displacement probability termed fractional anisotropy (FA). Because local tissue architecture, cell bodies, and myelinated axon fibers present distinct constraints on the degree and direction of water diffusion, it is possible to extract a map of brainwide structural connectivity from a 3D voxel-by-voxel map of FA values.

Matters of the white matter

An increasing number of groups are already applying diffusion-weighted MRI techniques to study brain structure in schizophrenia patients. Several additions to this literature were found in poster sessions at last year’s SfN meeting. A group of Kyoto University researchers used DTI to compare thalamo-prefrontal connectivity in 37 patients with schizophrenia and 36 age-, gender-, and education-matched healthy controls (Kubota et al. SfN 2011, 566.08). In addition to finding reduced size of the thalamo-prefrontal white matter tract in the right hemisphere of schizophrenia patients, first author Manabu Kubota and colleagues found that tract size also correlated with measures of cortical thickness, specifically in the orbitofrontal cortex where these fibers project. This provides a link between prior MRI and postmortem findings of reduced frontal cortex and thalamic size, suggesting a tight relationship exists between the two pathologies. However, how this specific structural pathology relates to particular symptoms of schizophrenia, whether cortical thinning leads to decreased white matter tract size, or the reverse, or if the pathology is causal of schizophrenia or a byproduct of disease onset, all remain open questions.

Researchers from the University of California, Los Angeles, and the University of Queensland in Brisbane, Australia, used DTI to look for structural connectivity that may confer risk for schizophrenia (Ryles AB et al. SfN 2011, 680.21). In a cohort of 306 healthy twin adults carrying a schizophrenia risk allele—a polymorphism in the NTRK1 gene, which codes for a high-affinity receptor for nerve growth factor (NGF)—first author April Ryles and colleagues found that allele carriers had lower connectivity within medial temporal and superior temporal lobes of the left hemisphere. Launching from prior functional imaging findings detailing aberrant lateralized activity in the left hemisphere of schizophrenia patients, they measured DTI tract sizes within and between cortical hemispheres. They found that tract sizes were specifically reduced in left temporal lobe white matter regions, suggesting that deficits in structural connectivity in this area could represent an anatomical basis of increased genetic risk for schizophrenia.

Both of these studies offer interesting contributions to current disconnectivity literature. To date, structural and functional neuroimaging studies in schizophrenia patients have revealed trends towards reduced global brain connectivity, temporal-frontal lobe disconnectivity, and the frequent involvement of frontal cortical regions (reviewed in Pettersson-Yeo et al., 2011). However, it is important to note certain difficulties in interpreting DTI findings in schizophrenia patients.

Some caveats

The most evident limitation stems from issues of variability within and between healthy and schizophrenia populations. As various research groups may choose to study patients representing childhood-onset schizophrenia, chronic schizophrenia, or genetic risk for schizophrenia, differences in connectivity could vary due to the developmental stage of the disease or the individual. It is, therefore, important to place individual connectivity findings into proper subgroup contexts. A number of recent imaging studies have detailed dramatic changes in brainwide connectivity in children and aging adults (Fair et al., 2009; Davis et al., 2012). In addition, some connectivity findings in chronic schizophrenia patients may potentially represent rewiring caused by medication and not disease development itself. Finally, among groups with genetic risk for schizophrenia, it is important to consider the specificity of the risk and the likelihood of subsequently developing schizophrenia. If particular disconnectivity findings are common in at-risk groups who develop distinct psychiatric diseases, or who remain healthy throughout life, this could weaken the use of these findings as reliable biomarkers for predicting and diagnosing schizophrenia.

Arguably, the most significant drawback of diffusion-weighted MRI is the assumption that FA values are accurate reflections of white matter tracts. As white matter in the brain may take complex paths—twisting, fanning, and crossing other fibers—it is possible that FA values in certain brain areas are imperfect reflections of white matter anatomy. A major goal of the NIH-funded Human Connectome Project is to develop improved diffusion-weighted techniques that are capable of more detailed and accurate brain mapping. Although valuable, these new approaches will still be limited in their ability to measure quantitative, micro-scale metrics of white matter anatomy, such as axon diameter, total axon number, fiber density, and degree of myelination. Decreased white matter size observed with diffusion-weighted MRI could thus reflect different combinations of these anatomical deficits, which could alter underlying assumptions about how macro-scale alterations in white matter affect neural communication.

In addition, critics of both connectomics and structural studies in general accurately point out: structural connectivity does not share a direct relationship with functional connectivity. Indeed, while the astronomically simpler connectome of the flatworm C. elegans—comprising just 302 neurons and 8,000 synapses—was mapped in 1986, progress in understanding how the organism’s neural connectivity relates to network function and behavioral output has been slow. In her presidential lecture at last year’s SfN meeting, Cornelia Bargmann of Rockefeller University, New York, detailed the ways in which vastly different behaviors of the flatworm could be produced by the same underlying anatomy—a fact that is not evident from a simple connectivity map. However, as decoding the genome was a necessary first step, and now an invaluable tool, in understanding the relationship between genes and behavior, Bargmann argued that the connectome is necessary but not sufficient for understanding the relationship between neural connectivity and behavior.

Given that disconnectivity in schizophrenia could involve altered structural and/or functional connectivity, thorough tests of the hypothesis require unified measures and analyses of both neural structure and function. The growing field of human connectomics can aid this effort by continuing to establish and improve methodological approaches for the acquisition and analysis of structural brain maps, from which direct comparisons in clinical populations can be made. Additionally, advances in complementary neuroimaging measures of "functional connectivity," such as resting-state functional MRI (fMRI), can serve to bridge the gap between structure and function.—Scott Bolkan.

See also Part 2.