Perinatal stroke is a focal, vascular brain injury occurring between 20 weeks gestation and 28 days post birth (Raju et al., 2007). With an estimated prevalence of 1:1100 (Dunbar et al., 2020), perinatal stroke is a common disruptor of early neurological development. The two most common subtypes of perinatal stroke are arterial ischemic stroke (AIS), caused by cerebral artery occlusion, and periventricular venous infarction (PVI), caused by hemorrhage in the germinal matrix in utero (Dunbar & Kirton, 2018). In addition to frequently resulting in cerebral palsy and epilepsy, perinatal stroke can impact intelligence, executive function, and language (Kirton & deVeber, 2013; Lee et al., 2005; Murias et al., 2014). Cognitive performance following perinatal stroke varies broadly (Murias et al., 2014; Westmacott et al., 2010), and may be influenced by lesion size, location, and time of injury among other factors (Ballantyne et al., 2008; Westmacott et al., 2010). Children with combined cortical and subcortical lesions tend to exhibit the largest disruptions to cognitive function (Westmacott et al., 2010).
Perinatal stroke lesions are more common in the left hemisphere and most commonly caused by middle cerebral artery infarcts (AIS subtype), often directly impacting language areas (Ballantyne et al., 2008; Dunbar et al., 2020; Núñez et al., 2020; Stephan-Otto et al., 2017; Vargha-Khadem et al., 1991). However, the majority of children who experience perinatal stroke achieve near-normal language performance, with functional deficits observed in only 20–25% of cases (Anderson et al., 2011; Bates et al., 2001; Fuentes et al., 2016; Kirton & deVeber, 2013; Lee et al., 2005; Murias et al., 2014; Stiles et al., 2005). This highlights the extraordinary ability of the brain to reorganize during early development; similar lesions during later childhood or adulthood can result in severe and permanent loss of language function. Remarkably, although language is typically left-lateralized, lesion side has only mild effects on outcomes (Ballantyne et al., 2007; Kirton & deVeber, 2013; Murias et al., 2014; Staudt, 2002). Language deficits may include more morphological errors, lower mean length of utterance, less complex syntax, and less detailed story settings (Avila et al., 2010; Demir et al., 2010; François et al., 2021; Reilly et al., 2013). Basic language skill is often preserved, with deficits observed when higher-level functions are tested (Ballantyne et al., 2007, 2008; Lee et al., 2005; Northam et al., 2018; Reilly et al., 2013; Westmacott et al., 2010). This suggests the developing brain can compensate for early damage to language areas, although the result may not be as efficient as a typical network.
Two reorganizational patterns may support language outcomes following a perinatal lesion (François et al., 2021; Murias et al., 2014). First, functional MRI studies of unilateral perinatal stroke have demonstrated a shift in processing to homologous language regions of the undamaged contralesional hemisphere (François et al., 2016, 2019; Guzzetta et al., 2008; Ilves et al., 2014; Jacola et al., 2006; Lidzba, 2007; Raja Beharelle et al., 2010; Staudt, 2002; Szaflarski et al., 2014; Tillema et al., 2008). Second, the language network may recruit remaining tissue in the ipsilesional hemisphere. Increased involvement of additional ipsilesional regions such as the posterior superior temporal gyrus or perilesional tissue has been associated with better language outcomes (Raja Beharelle et al., 2010; Vias & Dick, 2017). These two patterns are not mutually exclusive, and an altered interhemispheric balance in functional activity may be adaptive (Raja Beharelle et al., 2010).
While functional reorganization of language following perinatal stroke has been investigated, few studies have explored white matter connections underlying the language network (François et al., 2016, 2019; Heller et al., 2005; Northam et al., 2018). These studies assessed volumetric tissue damage and functional language outcomes and showed structural damage to left-hemisphere language tracts may result in a rightward shift of language function. However, as tissue volume is a non-specific metric, the conclusions drawn from these works should be considered preliminary. Studies employing metrics sensitive to white matter microstructure, such as diffusion-weighted MRI sequences, may provide new insight into neuroplasticity as related to language function, but no such studies exist to date.
Diffusion tensor imaging (DTI) is a popular diffusion-weighted MRI sequence that provides measurements of fractional anisotropy (FA), and mean, axial, and radial diffusivity (MD, AD, RD). These metrics are broadly sensitive to white matter microstructural features including myelin, axonal packing, axon permeability, and fiber coherence (Beaulieu, 2002). Beyond DTI, neurite orientation dispersion and density imaging (NODDI) provides the neurite density index (NDI) and orientation dispersion index (ODI) which are specific to axonal packing and fiber coherence, respectively (Zhang et al., 2012). Studies of brain development employing DTI and NODDI metrics have established that FA and NDI tend to increase while MD, AD, RD, and ODI tend to decrease with age. These trends suggest increases in axonal packing, myelin, and fiber coherence across healthy development (Chang et al., 2015; Geeraert et al., 2019; Mah et al., 2017). DTI and NODDI metrics have also been applied in perinatal stroke to describe links between white matter and sensorimotor function (Craig et al., 2022; Hodge et al., 2017; Kuczynski et al., 2017, 2018; Mailleux et al., 2020; Nemanich et al., 2019).
Here, we applied DTI and NODDI to assess white matter microstructure of two primary language-related tracts, the arcuate fasciculus (AF) and uncinate fasciculus (UF), in children with unilateral perinatal stroke as compared to typically developing controls (TDC). We hypothesized that FA and NDI would be lower, while MD, AD, RD, and ODI would be higher in ipsilesional tissue, with the opposite trend observed in contralesional tissue. Furthermore, we hypothesized that perinatal stroke participants with higher FA and NDI and lower MD, AD, RD, and ODI in language-related tracts would perform better on tests of language function.