Abstract
Aim To determine associations between 2-year language outcomes and near-term equivalent age diffusion MRI (dMRI) white matter tract metrics in a cohort of preterm children.
Method 32 children, gestational age 28.9 (2.4) weeks, underwent dMRI at mean post-menstrual-age 36.4 (1.4) weeks and language assessment with the Bayley Scales of Infant Development–III at 18-24 months. Mean fractional anisotropy (FA) was calculated for 6 white matter tracts. Hierarchical linear regression assessed associations between tract FA, moderating variables, and language.
Results Language outcomes correlated positively with FA of the left inferior longitudinal fasciculus (ILF-L) (r = 0.44, p < 0.05), and negatively with FA of the posterior corpus callosum (r = −0.41, p < 0.05). Language-brain associations in the ILF-L were independent of the number of neonatal medical complications. Number of neonatal medical complications moderated the relationship between language and posterior corpus callosum FA.
Interpretation The findings suggest that microstructural properties of white matter tracts in the newborn period are associated with language outcomes in the toddler years. The direction of associations varied by fiber group and may relate to the stage of brain maturation and/or the nature and timing of medical complications related to preterm birth.
INTRODUCTION
Children born preterm (PT) are at risk for expressive and receptive language deficits (Barre, Morgan, Doyle, & Anderson, 2011; Foster-Cohen, Edgin, Champion, & Woodward, 2007; Spek, Franken, & Weisglas-Kuperus, 2012). Language deficits in PT children increase risk for later reading problems, (Wocadlo & Rieger, 2007) which, in turn, are associated with adolescent health behaviors, and long-term academic and occupational success (Connor, Alberto, Compton, & O’Connor, 2014; Sanders, Federico, Klass, Abrams, & Dreyer, 2009). Efforts to understand the neurobiological basis for early language development in PT children may suggest avenues of intervention to improve language skills.
PT infants are at risk for white matter alterations (Volpe, 2009). Between 24 and 32 weeks gestational age (GA), pre-oligodendrocytes, the immature progenitors of myelinating glial cells (ie.,oligodendrocytes), are susceptible to oxidative stress from hypoxia, ischemia, and inflammation, often failing to mature or myelinate properly (Volpe, 2009). Periventricular regions are particularly sensitive to oxidative stress-related injuries (Volpe, 2009). Diffusion MRI (dMRI) has become a preferred method for assessing and characterizing the properties of white matter pathways in the brain (Basser & Pierpaoli, 1996; Feldman, Yeatman, Lee, Barde, & Gaman-Bean, 2010). Common dMRI metrics used to index white matter microstructure are fractional anisotropy (FA) and mean diffusivity (MD). Within white matter voxels, increased FA and decreased MD are generally associated with favorable neurobiological factors, such as increased myelination and axonal density. However, dMRI metrics are also sensitive to additional structural and tissue properties that reduce FA and/or increase MD, including the proportion of crossing fibers within a voxel, axonal size, and membrane permeability (Jones & Cercignani, 2010). Evidence for lower FA and/or higher MD in dMRI studies of PT neonates performed at n-TEA compared with children born full term, (Ball et al., 2013; Thompson et al., 2011) is consistent with evidence for reduced myelination and in axonal growth in animal models of preterm birth (Volpe, 2009).
Studies have examined relations between language and white matter pathway metrics in school-aged PT and FT children, (Dodson et al., 2018; Travis, Ben-Shachar, Myall, & Feldman, 2016) but few studies have examined similar white matter at near term equivalent gestational age (n-TEA). Emerging evidence suggests that variability in white matter microstructure obtained with dMRI at n-TEA relates to later language development in PT children. FA in the left and right arcuate fasciculi measured at 39-46 weeks was found to be independently associated with language outcomes at 22 months (Salvan et al., 2017). In a longitudinal cohort of PT children imaged at n-TEA, 2- and 4-years of age, (Young et al., 2017) slower rate of change of mean and radial diffusivity in the left internal and external capsules, and slower rate of change of axial diffusivity in the left posterior thalamic radiation was associated with lower 4-year old language scores (Bach et al., 2014). Additional studies are required to confirm and expand on these associations.
Preterm birth-related illnesses, such as bronchopulmonary dysplasia, have also been associated with white matter metrics (Dubner et al., 2019; Shim et al., 2012) and with language outcomes (Singer et al., 2001). The degree to which PT children experience neonatal medical complications varies substantially. Studies have yet to determine whether medical complications of preterm birth, especially those likely to induce hypoxic, ischemia or neuroinflammatory responses, moderate associations between near-term measures of white matter microstructure and language outcomes. Examining these relations in detail is pertinent to determining whether white matter microstructure may be an important early biomarker for language outcomes.
The present study used advanced, anatomically-specific tractography methods to determine if language outcomes at two years related to variability in white matter microstructure across multiple language-related white matter tracts at n-TEA in PT children and whether the number of medical complications moderated any effects. We obtained measures for FA and MD from white matter pathways commonly implicated in processing linguistic information, including a dorsal pathway, the left arcuate fasciculus (Arc-L) and two ventral pathways, the left and right inferior longitudinal fasciculi (ILF-L and ILF-R) and the left and right uncinate fasciculi (UF-L, UF-R) (Hickok & Poeppel, 2007). Additionally, we examined fiber pathways traversing the occipital segment of the corpus callosum (CC-Occ) to capture regions susceptible to injuries from medical complications of PT birth (Dubner et al., 2019; Thompson et al., 2011; Travis et al., 2019). We hypothesized that language would be associated with white matter metrics in these tracts, independent of post menstrual age at scan (PMA). In addition, we hypothesized that the number of neonatal medical complications would moderate the association in posterior white matter tracts because of their putative susceptibility to oxidative-stress related injuries.
METHODS
Participants
Children ≤1500 grams and ≤32 weeks gestation at birth from the Lucile Packard Children’s Hospital Stanford Neonatal Intensive Care Unit born between January 2010 and December 2011 were eligible for this prospective study (Rose et al., 2015). Exclusion criteria included evidence of genetic disorders or congenital brain anomalies. 102 participants were recruited, representing 76% of eligible infants. Of this group, 66 had dMRI scans at n-TEA. Participants were excluded due to T1w scan issues (n = 9), incomplete diffusion scans (n = 1), excessive head movement (n = 21, see supplementary material), and missing language measures (n = 3). The final sample was 32 participants.
Clinical variables were obtained via chart review. Additional description of the participants and outcomes using different analytic strategies was previously reported (Rose et al., 2015). The experimental protocol was approved by the Stanford University Institutional Review Board #IRB-13899. A parent or legal guardian provided informed written consent and participants were compensated for participation in neurodevelopmental assessments.
Procedures
Diffusion MRI Acquisition, Measures, and Analyses
MRI data were acquired on a 3T Discovery MR750 scanner equipped with an 8-channel HD head coil (General Electric Healthcare, Little Chalfont, UK) at the end of routine n-TEA neuroimaging lasting approximately 25 minutes. Infants were swaddled, fed and typically remained asleep without sedation during the scan. Results of the conventional MRI scans were obtained by chart review. The diffusion-weighted sequence, consisting of a 25-drection diffusion-weighted scan (b = 1,000 s/mm2) with 3 b = 0 scans was repeated twice in all but 2 participants. The diffusion scan with the least motion was selected for analysis. Imaging parameters, data preprocessing steps, motion correction procedures, diffusion tensor estimation, and individual native space tractography are described in the Supplementary Material and previous publications (Dodson, Travis, Ben-Shachar, & Feldman, 2017; Travis, Adams, Kovachy, Ben-Shachar, & Feldman, 2017). We analyzed 2 diffusion tensor metrics: FA and MD. Deterministic tractography, fiber tract identification, segmentation and quantification were implemented using the Automated Fiber Quantification (AFQ; https://github.jyeatman/AFQ) software package (Yeatman, Dougherty, Myall, Wandell, & Feldman, 2012) and MATLAB (MATLAB and Statistics Toolbox Release R2014a, The MathWorks, Inc., Natick, Massachusetts, United States). White matter pathways for analysis were selected a priori based on evidence for associations with language skills (Hickok & Poeppel, 2007) (Arc-L, ILF-L and ILF-R, UF-L and UF-R) or vulnerability to injury in PT children (CC-Occ) as shown in Figure 1 (Dubner et al., 2019; Thompson et al., 2011; Travis et al., 2019). Tract renderings in each participant were visually inspected prior to statistical analyses to ensure that results of tract segmentation conformed with anatomical norms. Tract profiles of FA and MD were calculated at 30 equidistant locations along the central portion of each fiber tract bounded by the same two regions of interest used for tract segmentation. Tract profiles were averaged to produce a single mean value for each tract.
Neurodevelopmental Assessment
Participants attended the High-Risk Infant Follow-Up program at 18-24 months corrected age, during which they were assessed on the Bayley Scales of Infant Development, 3rd Edition (BSID-III) (Bayley, 2006). One participant had follow up at 29.5 months chronological age. Per clinic protocol, this participant’s BSID-III scores were not corrected for GA, because they were obtained at ≥24 months corrected age. Exclusion of this participant did not change the analysis results. We analyzed the Language composite standardized score and the Expressive and Receptive Language subtest scaled scores.
Medical Complications
To assess the contributions of number of medical complications to outcomes, we generated a composite variable, MedRisk, summing the presence of 6 common medical complications of preterm birth (intraventricular hemorrhage, bronchopulmonary dysplasia, necrotizing enterocolitis, blood culture proven sepsis, patent ductus arteriosus, and retinopathy of prematurity). Participants were grouped into low risk (0 risk factors), medium risk (1 risk factor), and high risk (≥2 risk factors).
Statistical Analyses
Statistical analyses were conducted using SPSS (version 25.0, IBM Corp., 2018). Statistical significance was set at p < 0.05. The Shapiro–Wilk test was used to assess whether clinical and neurobiological data were normally distributed. With the exception of PMA, data were normally distributed. We chose to do parametric tests for all associations. A False Discovery Rate of 5% was calculated for zero-order associations to account for multiple comparisons (Benjamini & Hochberg, 1995).
Demographic and Clinical Variables
T-tests were computed to evaluate language outcomes for differences based on sex. Pearson correlations were computed to evaluate associations between language outcomes or white matter microstructure metrics, GA and PMA. One-way ANOVA assessed whether language outcomes differed by MedRisk.
Associations between Language Outcomes and mean White Matter Microstructure Metrics
Pearson correlations were computed to assess zero-order associations between composite, receptive, and expressive language scores and mean FA or MD. We then conducted a series of hierarchical linear regression models to assess the contribution of FA or MD to composite language outcome. We restricted regression analyses to tracts with zero-order correlations with uncorrected p-values < 0.05. PMA was entered in the models to ensure results were related to white matter microstructure metrics rather than participant age at scan. We assessed the main effects of MedRisk and the white matter metrics and then determined if MedRisk group moderated the prediction of FA to language outcome.(Hayes, 2017) The variance inflation factor (VIF) was calculated to assess multicollinearity of each model. We considered VIF values less than 10 to indicate that there was no concern for multicollinearity (Field, 2013).
RESULTS
Participant Characteristics
Participant characteristics, neonatal medical conditions, and language outcomes are shown in Table 1. PMA and GA were weakly negatively correlated (r = −0.28, p = 0.12). Clinical MRI scans demonstrated no abnormality (n = 17) or minimal abnormality (n = 14), such as subependymal hemorrhage or mineralization, (e.g. sequelae of germinal matrix hemorrhage) with or without mild ventriculomegaly, and with or without mild IVH Grade 1 or 2. One participant had parenchymal injury. Exclusion of this participant did not change the amount of variance accounted for by the multivariate models.
Language outcomes at two years
Participants demonstrated below-average language composite scores with substantial variation. Language scores did not differ by sex (p > 0.5) or MedRisk (F(2, 29) = 0.45, p = 0.65) and were not significantly correlated with PMA at scan or GA (all p > 0.05).
White matter microstructure at n-TEA in white matter tracts
We identified tracts conforming to anatomical norms in a majority of participants except the Arc-L, in which the tract was found in 16 (50%) of participants. Table 2 lists tracts, number of participants in which each tract was identified, and mean (SD) FA and MD. FA of the ILF-L, ILF-R, and UF-R significantly positively correlated with PMA. MD of the ILF-R, UF-L, and UF-R significantly negatively correlated with PMA. Correlations between tract FA or MD and PMA are shown in Table 3.
Correlations between two-year language outcomes and n-TEA white matter microstructural metrics
Correlations between language scores and mean FA are shown in Table 4A. Language composite significantly positively correlated with FA of the ILF-L, and significantly negatively correlated with FA of the CC-Occ. A moderate, positive correlation with FA of the Arc-L was of comparable strength, but did not reach statistical significance (p = 0.11). Receptive language significantly positively correlated with FA of the ILF-L and moderate correlations did not reach significance with FA of the Arc-L (p = 0.25) and CC-Occ (p = 0.070). Expressive language significantly positively correlated with FA of the ILF-L and approached significance with the FA of the Arc-L (p = 0.091).
Correlations between language outcomes and mean MD are shown in Table 4B. Expressive language significantly positively correlated with MD of the Unc-L only.
Hierarchical multivariable linear regression models
Table 5 shows the results of multiple regression models predicting language outcome at age 2 from n-TEA FA of ILF-L and CC-Occ, beyond variation accounted for by PMA and MedRisk. FA of the ILF-L contributed significantly to language composite (Model C), almost doubling the variance accounted for by PMA and MedRisk (Model B), and accounting for over 40% of the variance. The interaction term was not significant (Model D), indicating that the relationship between language outcomes at age two and FA of the ILF-L was similar for children with low, medium, and high MedRisk (Figure 2A). FA of the CC-Occ also contributed significantly to the language composite (Model G). The interaction of CC-Occ and MedRisk significantly increased the variance accounted for to about 35% (Model H), indicating that the relation between language outcomes and n-TEA FA of the CC-Occ was found in the participants with high- and medium-, but not low-MedRisk (Figure 2B).
DISCUSSION
In this investigation of 2-year-old children, we evaluated relations between white matter microstructure metrics from diffusion MRI at near term equivalent age and later language outcomes, accounting for the moderating effect of number of neonatal medical complications. The major finding of this study was that at age 2, language development was related to near term mean microstructural metrics of white matter pathways in a tract belonging to the ventral (left inferior longitudinal fasciculus) language processing stream. The moderate correlation of language and a dorsal pathway, the left arcuate fasciculus, may not have reached statistical significance given the small sample size. In addition, we found that composite language outcomes significantly negatively correlated with a tract likely to be susceptible to injury from preterm birth related complications: the occipital segment of corpus callosum. FA of the ILF-L uniquely contributed to variance in language outcome, beyond the variance accounted for by clinical course as measured by number of medical risk factors. We identified a moderating effect of number of medical complications on the association between language and FA in CC-Occ, but not ILF-L.
The associations between 2-year language outcomes and ventral tracts, are in agreement with associations between FA of the ILF-L and language in older children born preterm (Feldman, Lee, Yeatman, & Yeom, 2012; Murner-Lavanchy et al., 2018). The associations may correspond to consistently reported deficits in receptive language skills in children born preterm compared with their full term counterparts (Barre et al., 2011). In early language development, the ventral pathways, including the ILF, are associated with mapping auditory speech sounds to meaning and simple syntactic structure processing (Hickok & Poeppel, 2007).
A novel finding was that language outcomes were negatively associated with n-TEA FA of the occipital segment of the corpus callosum, a white matter region prone to injury in children born preterm, and not traditionally associated with language (Thompson et al., 2011). The correlation may indicate that FA in posterior periventricular regions indexes general white matter health or injury. Alternatively, or in addition, at near term, the posterior corpus callosum may be more relevant to language learning than later in childhood. The observed pattern for negative correlations within the posterior callosal pathway is consistent with reported negative associations between FA and readings skills in older preterm children (Dubner et al., 2019) and individuals born full term (Dougherty et al., 2007; Frye et al., 2008; Hasan et al., 2012; Huber, Henriques, Owen, Rokem, & Yeatman, 2019; Odegard, Farris, Ring, McColl, & Black, 2009). Diffusion metrics including FA and MD index multiple tissue properties including but not limited to myelin (Basser & Pierpaoli, 1996; Jones & Cercignani, 2010). Thus, the observed white matter-language relations in the CC-Occ may also reflect variations in axonal properties, such as fiber densities, (Huber et al., 2019) fiber coherence or increased axonal diameter, that have been shown to occur in different proportions across regions of the corpus callosum (Aboitiz, Scheibel, Fisher, & Zaidel, 1992). Future studies employing MRI techniques more specific to myelin and to axonal properties are required to clarify the relations between tissue properties across regions of the corpus callosum and later outcomes in children born preterm.
FA of the ILF-L explained additional variance in language beyond number of medical complications. We speculate that the unique contribution of white matter microstructure and the absence of moderation by medical complications in this pathway may relate to the importance of the maturational state of oligodendrocyte precursors relative to the timing and location of oxidative-stress induced brain injuries. For example, prolonged myelination of the ILF (Dubois et al., 2014) may reduce the susceptibility of such regions to injury in the neonatal period. Meanwhile the number of medical complications moderated the tract-language association in the posterior corpus callosum, with greater contribution of FA of the CC-Occ to prediction of language in children with more complications. The posterior corpus callosum myelinates earlier (Dubois et al., 2014) and is in a region more prone to injury in the neonatal period. A child with more medical complications in the neonatal period may have greater injury to oligodendrocyte progenitors (Volpe, 2009) in the posterior periventricular region. Our findings suggest that preterm children with fewer medical complications may rely primarily on language-related white matter pathways similar to full term children. For children with more medical complications, white matter may be more widely affected and therefore regions beyond traditional language pathways may contribute to variation in language development.
Limitations of the study include a modest sample size. We used a stringent motion criteria that, while useful in research, may limit direct applicability to the clinical setting. We did not have information about other variables associated with language outcomes, such as socioeconomic status (SES). Children with more medical complications may have been born to mothers with lower SES, which may be the true moderator of the relation between occipital corpus callosum FA and language. We used a simplistic measure of neonatal illness to group participants. While conditions such as necrotizing enterocolitis, bronchopulmonary dysplasia, sepsis, and intraventricular hemorrhage contribute to hypoxia, ischemia, and inflammation, the exact mechanism by which these conditions lead to changes in white matter metrics is unknown. These major medical complications may directly affect white matter or they may be a proxy for other disturbances such as prolonged poor nutritional status. Medical complications may also be a marker for other environmental influences associated with altered white matter microstructure, such as increased number of painful procedures (Vinall et al., 2014). These related endogenous or exogenous insults may be the ultimate culprits affecting white matter development.
CONCLUSION
In summary, a sensitive tractography method identified relations between language skills and n-TEA dMRI metrics of a specific ventral language processing stream white matter tract independent of the number of neonatal medical complications. In addition, neonatal medical complications moderated the relations between white matter microstructure metrics in the posterior corpus callosum, suggesting that the white matter properties associated with language outcomes may vary by fiber group, and relate to location and stage of myelination at injury and at the time of scan. Further in-depth exploration of neonatal clinical experiences and white matter pathways will contribute to understanding brain-language associations in the aftermath of preterm birth. A nuanced understanding of white matter microstructure in the neonatal period may be necessary to identifying imaging biomarkers for early identification of children at risk for adverse language outcomes.
Data Availability
Anonymized data will be made available upon request to qualified investigators for purposes of replicating procedures and results from the corresponding author.
Acknowledgments
This work was supported by National Institutes of Health [NICHD grant RO1-HD069162], and Health Resources and Services Administration Maternal Child Health Bureau [T77MC09796] to Heidi M. Feldman, PI, the National Institutes of Health [5K99HD084749] and the 2014 Society for Developmental and Behavioral Pediatrics Young Investigator Award to Katherine E. Travis, PI, and the Stanford Maternal Child Health Research Institute Tashia and John Morgridge Endowed Postdoctoral Fellowship to Sarah E. Dubner. This research was also supported by the Chiesi Foundation, Parma Italy; NIH Clinical and Translational Science Award UL1RR025744 for the Stanford Center for Clinical and Translational Education and Research (Spectrum) and for Stanford Center for Clinical Informatics, Stanford Translational Research Integrated Database Environment (STRIDE); Lucile Packard Foundation for Children’s Health; NSF Graduate Research Fellowship grant no. DGE-1147470, and by the Mary Baracchi Research Fund, Lucile Packard Children’s Hospital at Stanford.