Meta-analyzing brain dysfunctions in dyslexic children and adults

Abstract

We examined the evidence from functional imaging studies for predominance of a phonological left temporo-parietal (TP) dysfunction in dyslexic children and predominance of a visual-orthographic left occipito-temporal (OT) dysfunction in dyslexic adults. Separate meta-analyses of 9 studies with children (age means: 9–11 years) and of 9 studies with adults (age means: 18–30 years) and statistical comparison of these meta-analytic maps did find support for a dysfunction of a left ventral OT region in both children and adults. The findings on a possible predominance of a left TP dysfunction in children were inconclusive. Contrary to expectation, underactivation in superior temporal regions was only found for adults, but not for children. For children, underactivation was found in bilateral inferior parietal regions, but this abnormality was no longer present when foci identified by higher dyslexic task-negative activation (i.e., deactivation in response to reading compared to baseline) were excluded. These meta-analytic results are consistent with recent findings speaking for an early engagement of left OT regions in reading development and for an early failure of such an engagement in dyslexia.

Introduction

A substantial number of children suffer from unexpected and severe difficulties in learning to read and to write and these difficulties tend to persist into adulthood. The dominant explanation links these difficulties to a preceding language impairment and specifically to a phonological deficit (e.g., Shaywitz and Shaywitz, 2005, Snowling, 2000, Vellutino and Fletcher, 2005). The phonological deficit explanation, advanced by psycholinguists in the 1970s, is now endorsed by the World Health Organization, the International Dyslexia Association, and the National Institute of Child Health of the USA. In developmental terms, the language-phonological deficit affects the emergence of phoneme awareness at the beginning of learning to read, which then compromises acquisition of phonological word reading based on serial grapheme–phoneme coding. This primary decoding difficulty then has a negative effect on the build-up of the orthographic word lexicon required for fast automatic word recognition and correct spelling. This cognitive developmental account also served as framework for functional neuroimaging studies of dyslexia. In this field, a kind of standard model has emerged, which links the primary phonological word decoding difficulty to a dysfunction of a left dorsal temporo-parietal (TP) reading circuit which is assumed to include the posterior aspect of the superior temporal gyrus (STG) and the supramarginal and angular gyri of the inferior parietal lobule (IPL). The secondary visual-orthographic difficulty was linked to a dysfunction of a left ventral occipito-temporal (OT) reading circuit including lateral extrastriate, fusiform, and inferior temporal regions. These two left posterior dysfunctions were assumed to result in compensatory overreliance on less efficient right posterior regions or frontal articulatory regions. This model, originally advanced by Pugh et al. (2000), was used and elaborated in a number of reviews (e.g., Démonet et al., 2004, McCandliss and Noble, 2003, Sandak et al., 2004, Schlaggar and McCandliss, 2007, Shaywitz and Shaywitz, 2005).

A first quantitative attempt to link the evidence from imaging studies to the postulated developmental progression from a phonological left TP dysfunction to a visual-orthographic left OT dysfunction was provided by McCandliss and Noble (2003). These authors classified each of 11 imaging studies as providing evidence (i.e., underactivation) for the dorsal and/or the ventral dysfunction. Ten of the 11 studies found underactivation in dorsal circuit regions (termed “perisylvian”) and 6 of these studies found additional underactivation in the ventral circuit. A single study found a ventral dysfunction only. No proper evaluation of the developmental progression from TP to OT regions was possible since there were only 3 studies with children.

A more advanced approach to summarize evidence from imaging studies became possible with quantitative meta-analysis which is based on the fact that the majority of functional imaging studies report foci of brain activity in terms of 3D (x, y, z) coordinates in standardized stereotactic space. Such meta-analyses rely on foci of underactivation or overactivation of the original studies. These foci are treated as centers of probability distributions, which are then combined to create a whole-brain statistical map in order to estimate the likelihood of underactivation or overactivation for each voxel. A first application of this method to the field of dyslexia was done by Maisog et al. (2008) who included 9 studies (all published before 2006). The most robust finding was underactivation of left ventral regions, but underactivation was also identified in left dorsal (STG, IPL) and left inferior frontal regions. A recent meta-analysis from our lab by Richlan et al. (2009) included nearly twice the number of studies, but provided similar findings with respect to posterior dorsal and ventral dysfunctions. A new finding by Richlan et al. (2009) was the differentiation in the left frontal cortex between underactivation in the inferior frontal gyrus (IFG) and overactivation in precentral regions. Furthermore, overactivation was found in bilateral subcortical structures presumably engaged together with precentral regions in silent articulatory processes.

The meta-analyses by Maisog et al. (2008) and Richlan et al. (2009) did not differentiate between studies with younger and older dyslexic readers because, at that time, there were only few studies which focused on dyslexic children. The majority of studies were done with dyslexic adults and a smaller number was done with adolescents. Several studies included participants from a broad age range. Therefore, the results of the meta-analyses cannot provide information on the mentioned developmental progression from a phonological left TP dysfunction to a visual-orthographic left OT dysfunction. Furthermore, one may be concerned that the mentioned functional abnormalities of the meta-analyses in both left phonological TP regions and in left visual-orthographic left OT regions may have been due to the heterogeneity of the studies with respect to age and reading competence. The present study attempted to address these issues by differentiating between studies with younger readers (age 9 to 11 years) and studies with adults (age 18 to 30 years). This became possible because, lately, four fMRI studies with children were published. Altogether we were able to include 9 studies with dyslexic children and 9 studies with adults. For clear separation of age groups, studies with adolescents were not included. Based on these studies, separate meta-analyses were done for children and adults. Furthermore, the maps for children and adults were directly compared to gain information on developmental changes. Following the mentioned developmental account, we expected that phonological left TP dysfunction will be predominantly identified in the meta-analysis of children studies and, conversely, that visual-orthographic ventral dysfunction will be dominant in the meta-analysis of adult studies. One may also expect that precentral and subcortical overactivation reflecting silent articulatory processes may be more characteristic for younger than older dyslexic readers.