Tracking brain development for early prediction of language and literacy abilities

Language development begins before birth. Multiple lines of evidence demonstrated that the brain architecture supporting auditory and language processing develop during prenatal and early postnatal years. In the first chapter of the present Ph.D. thesis, the anatomo-functional emergence of primordial auditory and language networks in the foetal brain is described. The reviewed literature suggests that the structural and functional development of the auditory pathway is almost completed by the onset of the third gestational trimester, whereas language-related brain regions, their structural and functional connectivity develop in the last weeks before birth term. As reviewed in the first chapter, the interplay between genes and environment uniquely drives brain developmental processes, establishing the foundation upon which future language and literacy abilities can emerge. Understanding the intricate relationship between genetic and environmental factors, brain development and language acquisition represents one of the hardest challenges in cognitive neuroscience. The research presented in this Ph.D. thesis constitutes an effort to shed some light on this topic through different approaches. In the study detailed in the second chapter, resting-state fMRI data were acquired in a sample of preterm and at-term born newborns. The preterm newborns underwent longitudinal follow-up assessments during childhood, focusing on language, socio-emotional, and cognitive abilities. We explored the association of perinatal functional connectivity brain measures with premature birth. Subsequently, we investigated whether these measures could predict neurodevelopmental outcomes during early childhood. The results provide initial evidence about the association between perinatal functional connectivity brain markers in auditory and language regions, and children's cognitive abilities over their first three years. However, preterm born newborns are not entirely representative of typical prenatal development, as premature birth and postnatal exposure to external stimuli can significantly influence maturational trajectories. To address these limitations, in the study presented in the third chapter, resting-state fMRI data were obtained in a sample of healthy foetuses during prenatal development. After birth, and more specifically between 1 and 3 years of age, the same children were assessed for their expressive and receptive language abilities. A newly developed questionnaire was administered to the children’s parents to examine the familial susceptibility for language disorders, and to retrospectively estimate the amount of linguistic exposure received by the foetuses during pregnancy. The multiple relationships between familial risk, prenatal exposure factors, prenatal functional connectivity, and postnatal language abilities were investigated. The results showed that familial risk and prenatal exposure factors were associated with prenatal functional connectivity between brain regions that constitute the primoridial auditory and language processing systems. Moreover, prenatal functional connectivity in these systems was linked to expressive language skills in early childhood. While these two studies offer valuable insight into early language acquisition mechanisms, they are also limited to the investigation of resting-state functional connectivity, disregarding structural brain measures. In the fourth chapter, we describe a pilot study evaluating a novel paradigm that we specifically designed to identify early multimodal brain markers for the longitudinal prediction of developmental dyslexia, based on structural and functional brain measures and cognitive data. The pilot study envisaged multimodal data collection in a sample of young adults with and without a diagnosis of dyslexia. Between-group differences were first investigated for each of the brain and cognitive measures. Univariate and multivariate analyses showed that the combination of multimodal brain and cognitive measures can offer deeper insights into the neurocognitive bases underlying the dyslexic disorder. On top of this evidence, a machine learning model was trained to classify dyslexic and control subjects based on multimodal neuroimaging data: again, the combination of structural and functional brain measures yielded greater performance than the individual modalities alone. Moreover, the individual classification scores in the all-modalities model correlated with subject’s reading proficiency. In the general discussion provided in the last chapter, we highlight the emergence of an overarching bridging line that connects together the multitude of experimental findings presented in this Ph.D. thesis, namely the importance of functional and structural integration within auditory-language and sensorimotor brain networks for prenatal and early postnatal linguistic and cognitive development. Moreover, the set of all experimental studies is referenced to highlight the necessity of adopting a multifactorial perspective in the study of language development and reading disorders.