Altered Recruitment of the Brain's Reading Network in Children with Autism

Background: Autism spectrum disorders (ASD) are characterized by language deficits, which often entail difficulties in reading comprehension. Reading consists of decoding (the ability to translate letters into spoken language) and comprehension (the ability to understand what spoken words mean). While a proportion of children with autism show intact decoding skills, many exhibit significant deficits in comprehension (Nation et al, 2006). Such deficits may underlie altered recruitment of the language association cortex. Only a few neuroimaging studies have examined reading comprehension and its underlying neurobiology in autism. 

Objectives: To examine the functional integrity of the brain’s reading network in children with ASD and to relate it to their neuropsychological profile.

Methods: Nineteen high-functioning children with ASD and 16 typically developing (TD) control children participated in this study. The participants were matched on age (8-13 years), gender, handedness, full-scale IQ, verbal IQ, and fluency/decoding. Children completed a word similarities task in the scanner, in which they read a set of four words and determined whether the fourth word was similar to the previous three words presented (e.g. orange, apple, mango, table). fMRI data collected on a 3T Siemens Allegra scanner were analyzed using SPM12. Brain activation in regions within a previously identified reading network (Koyama et al., 2011) was examined.  

Results: A 2-sample t-test revealed significantly decreased activity in ASD children, relative to TD children, in the left superior-temporal gyrus (LSTG), left supramarginal gyrus/angular gyrus (LSMG/AG), and middle occipital area (p<0.05, 300 voxels). ASD participants had significantly lower reading comprehension scores (M = 77.6, SD = 13.3) than TD controls (M = 99.7, SD = 14.1); t(33) = 4.74, p<.001, as measured by the Gray Oral Reading Test (GORT-4). In addition, the GORT comprehension scores predicted activation patterns in ASD participants in bilateral frontal and thalamic regions, a trend not seen in TD participants. Similarly, the accuracy of ASD children in the word similarities task was positively correlated with their GORT comprehension scores (r = 0.559, p <0.05). 

Conclusions: Reduced activation in LSTG and LSMG/AG in ASD children may suggest difficulty in phonological as well as semantic processing (Bigler et al, 2007; Buchsbaum, 2001; Hickok & Poeppel, 2007). Since fluency was relatively intact in our sample, it is possible that ASD subjects used alternate methods to decode the words (e.g. memorization of words, pattern recognition). The relationship between GORT comprehension and bilateral frontal and thalamic activation in ASD children in our study may suggest a moderating effect of the thalamus on language comprehension in ASD (Klostermann, Krugel, & Ehlen, 2013). Thus, the results of this study provide evidence for altered recruitment of language areas in ASD children with intact fluency accompanied by poor comprehension; this may underlie differences in phonological and semantic processing. These findings have important implications for developing neurobiologically informed and targeted reading interventions in ASD.