Tract-based spatial statistics (TBSS) of diffusion tensor imaging data in alcohol dependence: Abnormalities of the motivational neurocircuitry

Abstract

Previous diffusion tensor imaging (DTI) studies indicated microstructural disruption of white matter in alcohol dependence. To investigate the microstructure of primary neurocircuitry involved in alcohol use disorders, the present study used Tract-Based Spatial Statistics (TBSS) of DTI measures as well as probabilistic tractography. Eleven recovering alcoholics in their first week of abstinence from alcohol were compared with 10 light-drinking controls; diffusion measures were correlated with measures of neurocognition and drinking severity. Regions characterized by low fractional anisotropy and high mean diffusivity included cortico-striatal fibers and those in frontal white matter and limbic pathways. Greater diffusion abnormalities in sections of commissural fibers (inter-hemispheric connections) were associated with greater drinking severity, and lower fractional anisotropy measures in frontal and limbic fiber tracts correlated with lower visuospatial memory performance. These study findings provide direct evidence of compromised integrity of the motivational brain circuitry in alcohol use disorders. These abnormalities in fiber connections could be partially responsible for deficiencies in executive functions, behavioral regulation, and impulse control commonly described in alcohol dependence.

Introduction

Water diffusion in the brain is influenced by the local microstructure of the tissue. Magnetic resonance diffusion tensor imaging (DTI) is a quantitative non-invasive and objective method for assessing the integrity of major white matter (WM) fiber tracts indirectly via measurement of the directionality of water diffusion (fractional anisotropy, FA) (Pierpaoli and Basser, 1996). FA reflects axonal diameter, axonal density and fiber tract complexity (Beaulieu, 2002). Water diffusion or mobility in WM is mainly hindered by myelin sheaths and cell membranes (Beaulieu, 2002) so that its movement is influenced by level of myelination and axonal density. More specifically, reduced FA can be attributed to degradation of both myelin sheaths and axonal membranes (Werring et al., 2000, Pierpaoli et al., 2001), abnormalities of myelin with sparing of the axonal fibers (Gulani et al., 2001, Song et al., 2002), or reduced density of axonal fibers (Takahashi et al., 2002). Mean diffusivity (MD), which is the overall mean-squared displacement of water molecules restricted by organelles and membranes (Neil et al., 1998), reflects cellular density and extracellular volume (Gass et al., 2001, Sotak, 2004) and relates to the volume fraction of the interstitial space (Sotak, 2004). Thus, low FA and high MD in the same brain region indicate higher intracellular or extracellular fluid associated with less organized myelin and/or axonal structure (Pfefferbaum and Sullivan, 2005). The spatial orientation of fibers can be derived from the eigenvalues from the diffusion tensor, and parallel (λ₁) and perpendicular (λ₂ and λ₃) diffusion to the WM tract can be calculated; these are also referred to as axial diffusivity, λ_|| = λ₁, and radial diffusivity, λ_⊥ = (λ₂ + λ₃)/2.

The hydrophobic myelin sheath hinders water movement perpendicular to axonal tracts (λ_⊥), imposing a preferential diffusion direction parallel (λ_||) to the WM tracts (Klingberg et al., 1999, Harsan et al., 2006). Therefore, an increase of λ_⊥ is thought to signify increased space between fibers (suggesting demyelination or dysmyelination) (Song et al., 2003, Harsan et al., 2006), whereas a decrease in λ_|| suggests axonal injury (Ito et al., 2002, Lazar et al., 2003).

In alcohol use disorders (AUD = alcohol abuse or dependence), microstructural WM integrity is thought to be disrupted by both demyelination and axonal structural abnormalities (Pfefferbaum and Sullivan, 2002, Pfefferbaum and Sullivan, 2005, Pfefferbaum et al., 2006a, Pfefferbaum et al., 2006b). These changes can be permanent or transient (Carlen et al., 1984). Permanent injury is thought to involve neuronal death with secondary axonal (Wallerian) degeneration, which has been documented in superior frontal lobe, hypothalamus and cerebellum (for review, see Oscar-Berman, 2000, Sullivan, 2000). Morphological abnormalities in AUD include transient changes and may include shrinkage of neuronal soma, dendritic pruning, decreased axonal size and density as well as myelin changes, which are manifested on magnetic resonance (MR) neuroimaging as reductions in brain tissue volume, metabolite abnormalities and decreased cerebral blood flow (Sullivan, 2000). Volumetric abnormalities may show partial to substantial recovery during short-term or extended sobriety (e.g., Pfefferbaum et al., 1995, Gazdzinski et al., 2005, Yeh et al., 2007). However, AUD-related microstructural abnormalities have been observed in regions of WM that appear grossly normal on structural MRI (Sullivan and Pfefferbaum, 2003). For example, FA reductions were reported in the genu of the corpus callosum (cc) and the centrum semiovale (cs) that exhibited no gross lesions on standard structural images (Pfefferbaum et al., 2000, Pfefferbaum and Sullivan, 2002, Pfefferbaum and Sullivan, 2005, Pfefferbaum et al., 2006a, Pfefferbaum et al., 2007).

Recent research in AUD has focused on the brain reward circuitry, which is implicated in the development and maintenance of substance use disorders in general (Chambers et al., 2003, Sullivan et al., 2003, Kalivas and Volkow, 2005, Sullivan and Pfefferbaum, 2005, Sullivan, 2003, Bowirrat and Oscar-Berman, 2005). This network is composed of dopaminergic afferents, which originate in the ventral tegmental area of the midbrain and project to the nucleus accumbens, prefrontal cortex, limbic, and striatal structures, and of reciprocal GABAergic or glutamatergic circuits connecting the dorsal anterior frontal cortex, orbito-frontal cortex, anterior cingulate cortex, and ventral striatum.

Current DTI data analysis methods, which include voxelwise analysis, region-of-interest (ROI) analysis, and tractography, suffer from the low spatial precision of cross-subject image alignment, the residual misalignment due to group differences in ventricular size, and the lack of knowledge of the proper degree of spatial smoothing. Additionally, traditional linear affine transformations may result in geometric tract distortion upon coregistration of WM tracts between groups. Tract-Based Spatial Statistics (TBSS), which uses nonlinear image transformation, is a newly developed technique that combines the strength of both voxelwise and tractography-based analyses (Smith et al., 2006). In this report, we demonstrate the application of TBSS to the investigation of the voxelwise differences of diffusion measures between alcohol-dependent individuals (ALC) and non- or light-drinking controls (LD). The goals of this study were 1) to implement TBSS of DTI data in ALC, 2) to describe and measure potential microstructural differences in the motivational neurocircuitry of ALC and LD, and 3) to correlate measures of drinking severity and neurocognition to the DTI measures of local WM microstructural abnormalities in ALC.