Quantitative MRI in the very preterm brain: Assessing tissue organization and myelination using magnetization transfer, diffusion tensor and T₁ imaging

Abstract

Magnetization transfer ratio (MTR), diffusion tensor imaging (DTI) parameters and T₁ relaxometry values were used to create parametric maps characterizing the tissue microstructure of the neonatal brain in infants born very premature (24–32 gestational weeks) and scanned at preterm and term equivalent age. Group-wise image registration was used to determine anatomical correspondence between individual scans and the pooled parametric data at the preterm and term ages. These parametric maps showed distinct contrasts whose interrelations varied across brain regions and between the preterm and term period. Discrete patterns of regional variation were observed for the different quantitative parameters, providing evidence that MRI is sensitive to multiple independent aspects of brain maturation. MTR values showed a marked change in the pattern of regional variation at term equivalent age compared to the preterm period such that the ordinal ranking of regions by signal contrast changed. This was unlike all other parameters where the regional ranking was preserved at the two time points. Interpreting the data in terms of myelination and structural organization, we report on the concordance with available histological data and demonstrate the value of quantitative MRI for tracking brain maturation over the neonatal period.

Introduction

About 2% of babies are born very premature, defined as under 32 weeks gestational age (Martin et al., 2010). This high risk population has over 50% morbidity rate (Wood et al., 2000) and is extremely susceptible to ischemic and hemorrhagic brain injuries, in white as well as gray matter areas of the brain (Barkovich, 2005, Rutherford, 2001). To better understand the impact of these injuries on brain development and obtain early predictors of outcome, a sensitive and specific means for monitoring brain development is needed.

Brain maturation is a sequential orchestrated process which includes the organization of the tissue into an elaborate and unique neuronal circuitry and the myelination of axons (Volpe, 2008). Studies of the human brain show these events begin in the second trimester of pregnancy and continue into adult life; the most rapid changes occurring from the 20th week of gestation to the end of the second postnatal year (Brody et al., 1987, Kinney et al., 1988, Yakovlev and Lecours, 1967). To date, MRI is the modality of choice for following normal as well as abnormal brain maturation in vivo, in particular myelination and sulcation (Barkovich, 2005, Rutherford, 2001). In vitro and in vivo MRI can demonstrate the establishment of proper alignment, orientation and layering of neurons as early as the fetal (below 26 weeks gestational age (GA)) and early preterm (26–34 weeks GA) periods (Brisse et al., 1997, Girard et al., 1995, Huang et al., 2006, Kostovic and Jovanov-Milosevic, 2006, Maas et al., 2004, Rados et al., 2006). The application of MR techniques such as magnetization transfer imaging (MTI), diffusion tensor imaging (DTI) and T₁ relaxometry enables non-invasive quantification of brain maturation by characterizing cerebral regional differences and providing insights into the underlying microstructure of the developing brain (Deoni et al., 2011, Engelbrecht et al., 1998, Hasan et al., 2010, Hermoye et al., 2006, Klingberg et al., 1999, Lebel et al., 2008, Mukherjee and McKinstry, 2006, Schneider et al., 2004, van Buchem et al., 2001, Xydis et al., 2006, Yoo et al., 2005).

These techniques yield different contrast parameters, namely, magnetization transfer ratio (MTR), DTI parameters (fractional anisotropy and mean, axial and radial diffusivity) and T₁ values, that are sensitive (but not specific) to water content in tissue, the progression of tissue organization, and early myelination events of the immature brain. All of these measures exhibit age-dependent asymptotic behaviour that is consistent with post-mortem measurements of water and lipid concentrations in the developing brain (Dobbing and Sands, 1973, Kinney et al., 1994). Nevertheless, distinct underlying biophysical variables govern these contrast mechanisms. Hence, a multi-modal approach combining findings of different MRI techniques may provide a more complete picture of brain maturation.

In the early preterm period the brain is about 92% water (Dobbing and Sands, 1973) and largely unmyelinated (Brody et al., 1987, Kinney et al., 1988, Yakovlev and Lecours, 1967). As maturation progresses, cellular density, axonal density, and the degree of coherent axonal alignment all increase, while the water content in tissue declines to 88% at birth. In parallel, pre-myelination events and early myelination processes begin, with the development of oligodendrocytes and premyelin lipids, resulting in higher concentration of myelin-related macromolecules (cholesterol, myelin basic protein and proteolipid protein, and lipids such as sphingomyelin, sulfatides and cerebrosides). As a consequence, there is more restriction and hindrance in tissue causing increased water directionality (Volpe, 2008).

Previous in vitro magnetization transfer studies (Ceckler et al., 1992, Fralix et al., 1991, Koenig, 1991, Kucharczyk et al., 1994) and in vivo work in early stages of brain development (Engelbrecht et al., 1998) have ascribed magnetization transfer to the presence of macromolecules associated with myelination, in particular cholesterol and glactocerebrisides. Although at later stages of brain maturation the increasing concentration of these macromolecules has been also connected to T₁ shortening (Koenig, 1991), in the preterm period T₁ values have been mostly related to cellular density (Girard et al., 1995) and free and total tissue water content due to the relatively low concentration of semisolids in tissue (Williams et al., 2005). MTRs provide a measure of magnetization transfer in tissue and by extension myelination, but are sensitive in varying degrees to multiple other factors including concentrations of other macromolecules, magnetization exchange rate, water concentration, T₁ and T₂. Whereas biophysical models predict that MTR should increase with semi-solid fraction and to a lesser extent T₁ (Henkelman et al., 2001), we observed a negative correlation between MTR and T₁ in deep grey matter (GM) structures in the very preterm brain (Nossin-Manor et al., 2012), emphasizing that MTR and T₁ are distinct measures of tissue microstructure at early stages of maturation.

We have previously reported higher MTR and lower T₁ values measured in the very preterm brain in early myelinated deep GM structures, such as the pons and thalami, compared to unmyelinated GM structures, such as the basal ganglia (BG) (Nossin-Manor et al., 2012). Previous MTR measurements in children aged 1 week to 6 years showed that at birth the highest MTR values were measured in early myelinated GM and white matter (WM) regions such as the midbrain, the posterior pons and the middle cerebellar peduncle, whereas the lowest values were found in unmyelinated structures such as the frontal and occipital WM (Engelbrecht et al., 1998). A more recent study in prematurely born infants at term showed the highest MTR values in the splenium of the corpus callosum (CC), a WM structure not showing mature myelin until 12 weeks after term, and the thalamus, a partly myelinated deep GM structure (Xydis et al., 2006). By 2 years, both studies demonstrated changes in regional order and showed the highest values in the CC, presenting a high number of heavily myelinated fibers, and the lowest in deep GM structures.

T₁ relaxometry measurements reported for neonates at term, preterm infants at term equivalent age and for preterm brain at birth showed lower T₁ values in GM structures compared to non-myelinated WM (Jones et al., 2004, Nossin-Manor et al., 2011, Williams et al., 2005). Furthermore, a study in healthy term-born infants from 3 to 11 months of age demonstrated lower T₁ values in GM structures compared to unmyelinated WM and in myelinated WM compared to GM structures (Deoni et al., 2011).

DTI parameters reflect restriction of water movement in tissue (Drobyshevsky et al., 2005, Huppi et al., 1998, Le Bihan, 2003, Neil et al., 2002): Fractional anisotropy (FA) is a composite measure of the extent of tissue directionality (Pierpaoli and Basser, 1996). Mean diffusivity (MD) reflects hindrance and restriction of water movement in tissue (Le Bihan, 2003). Axial diffusivity (AD) reflects axonal alignment and coherent orientation organization, while radial diffusivity (RD) indicates restriction perpendicular to the axon orientation (Song et al., 2002). DTI studies investigating WM as well as GM regions in the fetus, preterm and full-term newborn brain have demonstrated a defined regional hierarchy using measurements of FA MD, AD and RD (Berman et al., 2005, Dubois et al., 2006, Huppi et al., 1998, Kasprian et al., 2008, Neil et al., 1998, Partridge et al., 2005). This hierarchy, however, does not fully correspond to the regional sequence of myelination presented by histology (Brody et al., 1987, Kinney et al., 1988, Yakovlev and Lecours, 1967). For example, while the highest FA values were found in the splenium followed by the genu of the CC and posterior limb of the internal capsule (PLIC), myelination occurs at around 36 weeks GA in the PLIC and only at about 12–20 weeks after birth in the CC. Moreover, similar regional differences in anisotropy indices have been reported in myelinated fibers in children and young adults (Lebel et al., 2008, Mukherjee et al., 2001, Zhai et al., 2003). A study in healthy term-born infants in their first months of life investigating WM using DTI and tractography showed similar organization to adults (Dubois et al., 2006). While normalization of DTI metrics against regional values for adults and term age infants (Dubois et al., 2008a) provides a means to stage myelination in the presence of these large regional variations, it is clear that myelination is not the primary factor governing DTI measures in the developing brain. Previous animal studies support this view, by showing only 10–15% reduction in anisotropy values in the absence of myelin (Beaulieu and Allen, 1994a, Song et al., 2002). Early brain development studies in humans (Hermoye et al., 2006, Huppi et al., 1998, Neil et al., 1998), rabbits (Drobyshevsky et al., 2005) and rodents (Wimberger et al., 1995, Zhang et al., 2003) provide additional supporting evidence showing high anisotropy in the pre-myelinated WM followed by a 25–30% increase upon myelination (Zhang et al., 2003); the latter increase being coupled to histological verification of mature myelin (Wimberger et al., 1995). Other tissue features modulating anisotropy are axonal membrane permeability (Beaulieu and Allen, 1994a, Beaulieu and Allen, 1994b), ion exchange (Prayer et al., 2001), and axonal distribution and density (Beaulieu and Allen, 1994a, McGraw et al., 2002). The complementary contribution of axonal organization and myelination has been proposed to explain changes in diffusion anisotropy values during development (Klingberg et al., 1999, Neil et al., 1998). Nevertheless, the integration of tissue organization and myelination in the developing brain has not been quantitatively addressed by MRI. Thus, a detailed analytical approach is needed to closely relate changes in quantitative MRI (qMRI) parameters with the processes shaping tissue microstructure during maturation.

Here, we present a multi-modal approach using combined MTR, DTI and T₁ measurements to follow brain maturation in very preterm neonates (born ≤ 32 weeks GA) at birth and at term equivalent age (TEA) and to quantitatively investigate the underlying microstructure of the developing brain by measuring regional variations over time. Dissociating these measurements, we hope to gain a more complete understanding of the components implicated in brain maturation and the timing of these developmental changes using in vivo MRI.