Atypical Structural Asymmetry of the Planum Temporale is Related to Family History of Dyslexia

Abstract

Research on the neural correlates of developmental dyslexia indicates atypical anatomical lateralization of the planum temporale, a higher-order cortical auditory region. Yet whether this atypical lateralization precedes reading acquisition and is related to a familial risk for dyslexia is not currently known. In this study, we address these questions in 2 separate cohorts of young children and adolescents with and without a familial risk for dyslexia. Planum temporale surface area was manually labeled bilaterally, on the T1-weighted MR brain images of 54 pre-readers (mean age: 6.2 years, SD: 3.2 months; 33 males) and 28 adolescents (mean age: 14.7 years, SD: 3.3 months; 11 males). Half of the pre-readers and adolescents had a familial risk for dyslexia. In both pre-readers and adolescents, group comparisons of left and right planum temporale surface area showed a significant interaction between hemisphere and family history of dyslexia, with participants who had no family risk for dyslexia showing greater leftward asymmetry of the planum temporale. This effect was confirmed when analyses were restricted to normal reading participants. Altered planum temporale asymmetry thus seems to be related to family history of dyslexia.

Introduction

The planum temporale is a cortical region that is functionally involved in higher-order auditory and language processing (Price 2010). This triangular-shaped area located on the superior surface of the temporal lobe has been suggested to sustain phonological representations (Jäncke et al. 2002; Vigneau et al. 2006; Dehaene et al. 2010).

The asymmetry of planum temporale surface area varies along a continuum in the general population that is skewed to the left (e.g., Geschwind and Levitsky 1968). In normal reading adults, 60–70% of the population shows a leftward asymmetry of the planum temporale (Geschwind and Levitsky 1968; Witelson and Pallie 1973; Wada et al. 1975; Galaburda et al. 1987; Chance et al. 2006). This structural lateralization towards the left hemisphere is also present in normal reading children (Hynd et al. 1990; Larsen et al. 1990; Semrud-Clikeman et al. 1991; Dalby et al. 1998; Bonte et al. 2013; Sanchez-Bloom et al. 2013; Altarelli et al. 2014), as well as in foetuses and infants (Witelson and Pallie 1973; Wada et al. 1975; Chi et al. 1977; Galaburda and Geschwind 1981). In addition to surface area, planum temporale gray matter volume and cortical thickness have also been measured. Leftward asymmetry of the planum temporale volume has been observed in adults (Meyer et al. 2014) while no asymmetry has been reported in cortical thickness (Altarelli et al. 2014; Meyer et al. 2014). Surrounding the planum temporale, white matter organization in the posterior part of the superior temporal gyrus (pSTG) has also been shown to be leftward lateralized in adults (Vandermosten et al. 2013). This asymmetry in pSTG white matter organization, quantified by fractional anisotropy, is likely to be important for spectrotemporal acoustic analysis (Hickok and Poeppel 2007; Lebel and Beaulieu 2009; Vandermosten et al. 2013).

Leftward asymmetry of the planum temporale was initially considered as a possible structural substrate of the left hemispheric functional dominance for language (Geschwind and Galaburda 1987). However, studies pairing analyses of functional and anatomical asymmetries in this region only found partial support for this hypothesis (Dorsaint-Pierre et al. 2006; Dos Santos Sequeira et al. 2006; Eckert et al. 2006; Keller et al. 2011). Functional dominance for reading has also been shown in left temporoparietal cortex, which includes the planum temporale, as well as in left occipitotemporal cortex and inferior frontal cortex (e.g., Pugh et al. 2000; Vigneau et al. 2006; Schlaggar and McCandliss 2007). The aforementioned left temporoparietal region is thought to link orthographic and phonological processes and sustains the graphophonological reading route (Simos et al. 2002; Jobard et al. 2003; van Atteveldt et al. 2004; Vigneau et al. 2006; Schlaggar and McCandliss 2007). The occipitotemporal region is involved in direct coupling of the visual word form to semantics and becomes dominant in more advanced reading stages, when grapheme–phoneme coupling is automatized (Jobard et al. 2003; Schlaggar and McCandliss 2007; Vandermosten et al. 2012).

Due to its neuroanatomical proximity to the primary auditory cortex, and given that it is part of a wider network of regions specialized for speech and language, the planum temporale has been considered as a prime candidate for the phonemic processing that accompanies reading acquisition (Turkeltaub et al. 2003; Dehaene et al. 2010; Golestani et al. 2011). In dyslexia, which is a learning disability characterized by severe and persistent reading and/or spelling impairments in the absence of intellectual and sensory deficits (Shaywitz 1998; Vellutino et al. 2004), phonemic processing deficits are considered as an underlying cause for the reading problems (Snowling 2000). Therefore, structural anomalies in the planum temporale may be among the main risk factors potentially leading to the emergence of this specific learning disorder.

Research on the structural neural substrates of developmental dyslexia has indeed focused on the planum temporale, with both “post mortem” (Galaburda et al. 1985; Galaburda 1989; Humphreys et al. 1990) and in vivo investigations (see Altarelli et al. 2014 for a summary table of all in vivo studies). Yet, results have been inconsistent. On the one hand, an altered anatomical lateralization pattern has repeatedly been observed in individuals with developmental dyslexia (Hynd et al. 1990; Larsen et al. 1990; Semrud-Clikeman et al. 1991; Hugdahl et al. 1998; Sanchez-Bloom et al. 2013; Altarelli et al. 2014), showing a larger proportion of rightward asymmetrical cases. On the other hand, a number of studies failed to reproduce these results in children (Schultz et al. 1994; Heiervang et al. 2000; Foster et al. 2002; Eckert et al. 2003; Hugdahl et al. 2003; Kibby et al. 2004) as well as in adults (Rumsey et al. 1997; Best and Demb 1999; Robichon et al. 2000). Some of these studies did, however, report a smaller left planum temporale in individuals with dyslexia (Rumsey et al. 1997; Heiervang et al. 2000; Hugdahl et al. 2003). Regarding the white matter connectivity literature, recent evidence suggests that white matter in the posterior part of the STG is more symmetric in adults with dyslexia compared with normal readers (NR) (Vandermosten et al. 2013). Finally, reduced activation in the planum temporale has been observed in 9- to 10-year-old children with dyslexia for processing speech sounds and mapping phonemes to graphemes (Blau et al. 2010)—yet no direct information on functional asymmetry patterns is available in that study.

It should be noted that confounding variables such as handedness and IQ have not been controlled for in the majority of anatomical studies on planum temporale, although these variables appear to be related to its asymmetry (Steinmetz et al. 1991; Shapleske et al. 1999; Eckert and Leonard 2000). The importance of sex in this context has recently been demonstrated by a study reporting an altered pattern of planum lateralization in dyslexic boys only (Altarelli et al. 2014). The labeling criteria and method used to delineate the planum temporale on MR images are also of crucial importance and have not been consistent across all studies (Shapleske et al. 1999; see Altarelli et al. 2014 for a summary of the methods used). Among the various existing labeling techniques, the technique proposed by Steinmetz et al. (1989) has been adopted and adjusted in many studies (e.g., Steinmetz et al. 1991; Hugdahl et al. 1998), in combination with the anatomical borders defined by Shapleske et al. (1999). The vast majority of studies measured the planum temporale in the sagittal plane only (e.g., Hynd et al. 1990; Semrud-Clikeman et al. 1991; Leonard et al. 1993; Hugdahl et al. 1998; Robichon et al. 2000; Foster et al. 2002; Eckert et al. 2003; Leonard et al. 2006; Sanchez-Bloom et al. 2013). However, it has been argued that delineation of the planum temporale based on sagittal slices causes an underestimation of its lateral extension (Kulynych et al. 1993; Loftus et al. 1993) and thereby results in a reduced measurement of the planum temporale. Simultaneous views of the 3 orthogonal planes form an advantage of crucial importance for accurate labeling (Galaburda 1993; Kulynych et al. 1993; Honeycutt et al. 2000). Throughout the years, imaging protocols have also evolved. While some, in general older, studies acquired data on a 0.5–0.6 T scanner (e.g., Hynd et al. 1990; Semrud-Clikeman et al. 1991; Foster et al. 2002), other studies used a higher quality 1.5 T MRI scanner (e.g., Hugdahl et al. 1998; Sanchez-Bloom et al. 2013), and fewer studies used a 3T MRI scanner (for an overview see Altarelli et al. 2014). The applied protocols also vary widely in slice thickness, ranging from 1 to 7.5 mm. Overall, inconsistencies arising from the comparison between studies using different labeling criteria and techniques are difficult to interpret and should be considered with caution.

Next to confounding effects that have been suggested by previous research, results on the lateralization of the planum temporale might be influenced by another variable that has been inconsistently used for participant selection. Indeed, as developmental dyslexia co-occurs in families (Gilger et al. 1992), an important factor to consider is the family history of dyslexia. Some studies used family history as a criterion for selecting individuals with dyslexia (Leonard et al. 1993; Robichon et al. 2000; Eckert et al. 2003; Sanchez-Bloom et al. 2013); in other studies, individuals with a family risk for dyslexia have also been included in the control group (Leonard et al. 1993; Sanchez-Bloom et al. 2013); and yet other studies did not report on the family risk of the control group at all (Robichon et al. 2000; Eckert et al. 2003). It appears that taking a more systematic approach to address this aspect is of importance.

The suggested atypical lateralization pattern reported by some studies in dyslexic individuals raises the question of its origin. What might be causing an altered anatomical lateralization in dyslexia? The commonly observed leftward pattern of planum temporale asymmetry in the general population might be resulting from greater neurodevelopmental pruning of the right side during foetal life and infancy (Geschwind and Levitsky 1968; Galaburda et al. 1987). Consequently, reduced leftward asymmetry in individuals with dyslexia might be due to insufficient pruning, resulting in an enlarged right planum temporale (Galaburda et al. 1987; Humphreys et al. 1990; but see Altarelli et al. 2014). Galaburda et al. (1987) suggested that a putatively enlarged right planum temporale might be related to foetal testosterone. The effect of prenatal factors on planum temporale asymmetry is supported by studies on twins (Steinmetz et al. 1995; Leonard et al. 2006), which suggest a potential impact of intrauterine events on planum temporale length.

In addition to influences of prenatal environmental factors, planum temporale lateralization might be a partly inherited feature, as is the case for hand preference and developmental dyslexia. A putatively inherited component of planum temporale lateralization has been suggested in NR by Steinmetz and Galaburda (1991), who detected reduced leftward asymmetry in a subsample of left-handed individuals with first-degree left-handed relatives (familial sinistrality). The hypothesis of a genetic influence on brain asymmetries has also found partial support in more recent work (for a review see Brandler and Paracchini 2014).

This study is the first to directly investigate PT lateralization in individuals with and without a family history of dyslexia. We thereby aim to investigate whether the previously reported atypical asymmetry of planum temporale in the dyslexic population is related to a familial risk for the disorder, and whether this can clarify the previously reported conflicting results. Accordingly, we applied detailed anatomical analyses of the planum temporale, and measured planum temporale surface area in kindergartners (n = 54) and in adolescents (n = 28). In both cohorts, half of the participants had a familial risk for dyslexia (FRD⁺). Based on longitudinal cognitive data in each cohort, 12 kindergartners and 12 adolescents were (retrospectively) classified as dyslexic. While our study was not specifically designed to investigate the confounding influence of sex, handedness, and IQ, we controlled for these covariates in all analyses.

Method

Participants

Pre-readers were recruited from the longitudinal study by Vanvooren et al. (2014) in which originally 87 Dutch-speaking children participated. Half of them had a family risk for dyslexia (FRD⁺), defined as having at least one first-degree relative with a formal diagnosis of dyslexia, while the other half had no family risk for dyslexia (FRD⁻). For more information on participant selection criteria see Vanvooren et al. (2014) and Vanderauwera et al. (2015). Of the original sample, 75 pre-readers participated in the MRI study and T1 scans were administered in 73 children at the end of the last year of kindergarten. Data of 4 children were excluded because of unsuccessful scan acquisition and 8 children were excluded because of extensive motion, based on visual inspection of the data. For this study, 5 criteria were used, ensuring that all participants had 1) a normal non-verbal IQ, that is, a standardized score ≥80 on the WISC-III-NL subtest Block Design measured at the start of grade 2 (Table 1) (Kort 2005), 2) no history of brain damage, 3) no history of psychiatric disorders, 4) no known visual deficits, 5) no left-handedness. For the fifth criterion, we administered the Edinburgh Handedness Inventory (Oldfield 1971), and participants were classified as left-handed (score ≤−40), right-handed (score ≥40), or ambidextrous (score between −40 and 40). Since a confounding effect of handedness on planum temporale lateralization has been suggested (Steinmetz et al. 1991; Shapleske et al. 1999; Eckert and Leonard 2000), left-handed pre-readers (n = 7) were excluded from all analyses. All participants had bilateral normal hearing, that is, a pure-tone average (PTA, defined as the mean threshold at 500, 1000 and 2000 Hz) <25 dB HL. As a result of the aforementioned selection criteria, 54 (33 boys/21 girls) 5- to 6-year-old children were included in this study. Half of the children (n = 27) had a familial risk for dyslexia (FRD⁺). The pre-readers were also retrospectively classified as dyslexic (DR) or NR based on standardized word reading, pseudoword reading and spelling tests administered at the start of grades 2 and 3 (Brus and Voeten 1973; Van den Bos et al. 1994). For inclusion in the DR group, children had to meet 1 of 2 criteria. The first criterion stated that the reading problems had to be persistent, that is, present at both time points, and severe, that is, participants had to score lower than or equal to percentile 10 on both measurements of one reading test. The second criterion selected participants with severe and profound spelling problems, that is, below percentile 10 on both measurements. For this second criterion, as dyslexia is mostly defined as a reading disorder, these participants were additionally required to score below percentile 25 at both measurement times on a reading test. Twelve of the 54 pre-readers included in this study developed dyslexia (10 FRD⁺, 2 FRD⁻). The children that participated in this study did not receive formal reading and writing instruction in kindergarten, in line with the guidelines of the Flemish government (http://www.ond.vlaanderen.be/).

Adolescents were recruited from the longitudinal study of Boets et al. (2011), in which 62 participants were included. Similar to the pre-readers, half of the original participants had a familial risk for dyslexia. This study reports on neurological measures gathered in adolescence, while the longitudinal study of Boets et al. explored cognitive markers of dyslexia prior to reading onset and throughout reading acquisition. For more details on participant selection criteria and on the tasks that were administered previously in this sample see Boets et al. (2010, 2011). Of the original sample, 34 adolescents (age range 14–15 years) accepted to participate in this study. The same 5 inclusion criteria were applied as in pre-readers. Five subjects were excluded because they were left-handed. All participants had bilateral normal hearing, except 2 participants with a mild conductive hearing loss in the right ear (but still PTA <35 dB HL). Excluding these participants from all analyses did not change the results. All adolescents underwent a structural MRI scan and behavioral assessment battery. Due to dental braces, MRI data of one participant could not be used. In sum, 28 participants (11 boys/17 girls) met all criteria and were included in the analyses presented here. Standardized word reading, pseudoword reading test and spelling tests (Brus and Voeten 1973; Van den Bos et al. 1994) were conducted at grade 6 (age: 11–12 y) (Poelmans et al. 2011) and at the time of MRI acquisition (grade 9, age: 14–15 y). The criteria applied for the classification are identical to those applied in pre-readers. Twelve adolescents met these criteria and were included in the DR group (10 FRD⁺, 2 FRD⁻). This study was approved by the ethics committee at the University Hospital of Leuven. The parents of the participants gave their written consent for the participation of the children in the MRI scan and behavioral assessment battery, in line with the Declaration of Helsinki. Additionally, the adolescents provided written assent for participation witnessed by their parents.

MRI Acquisition and Analysis

A 3T MRI scanner (Philips, Best, the Netherlands) was used to acquire a structural T1 scan with a 32-channel head coil. In the pre-readers, 182 contiguous coronal T1 slices were collected using the following parameters: repetition time (TR) = 9.7 ms, echo time (TE) = 4.61 ms, flip angle = 8°, voxel size 0.98 × 0.98 × 1.20 mm. The scan acquisition time was 6:22 min. In adolescents, 2 averages were obtained of 190 contiguous sagittal slices, using the following parameters: TR = 7.0 ms, TE = 4.0 ms, flip angle = 8°, voxel size 1.0 × 1.0 × 1.0 mm. Total scan duration was 10:23 min.

The software BrainVISA (brainvisa.info) was applied to create for each individual a left and right cortical surface reconstruction (“mesh”). T1 images and corresponding gray matter surface meshes were aligned to AC-PC space (anterior commissure – posterior commissure).

Cortical Surface Measurement

The visualization and labeling tools of Anatomist software (Rivière et al. 2000; Le Troter et al. 2012) were applied for the labeling of the planum temporale. This software provides simultaneous views of the 3 orthogonal planes and of the gray matter surface meshes, on which the areas are labeled, which is an advantage of crucial importance for accurate labeling (Galaburda 1993; Kulynych et al. 1993; Honeycutt et al. 2000).

The planum temporale was delineated based on the highly detailed labeling method of Altarelli et al. (2014) (see Fig. 1). First, well-defined anatomical criteria were applied to label Heschl's gyrus (H1), following Altarelli et al. (2014). If multiple gyri of Heschl were present, the most anterior gyrus was defined as H1, and any additional Heschl's gyrus was defined as H2. When a posterior split of Heschl's gyrus was present, a distinction was made between full posterior duplications, dividing Heschl's gyrus into H1 and H2, and common-stem duplications, where the posterior split does not reach the medial retroinsular region. Based on myelogenetic and cytoarchitectonic work on these regions (Pfeifer 1920, 1936; Galaburda and Sanides 1980), and again consistently with Altarelli et al. (2014), H2 was considered a part of the planum temporale. In the present sample, it should be noted that no participant presented more than 2 gyri of Heschl.

Heschl's sulcus was considered as the anterior border of the planum temporale. The posterior border of the planum temporale was established on coronal and sagittal views, avoiding the inclusion of any irrelevant parietal sulcus, as thoroughly described by Altarelli et al. (2014).

The planum temporale surface was labeled on each hemisphere and for each participant, enabling the data to be processed in native space and thereby avoiding normalization artefacts. Cortical surface area was extracted for each label and for each hemisphere. Surface area measures were normalized by dividing the surface area of the planum temporale within each hemisphere by the corresponding mean hemispheric surface area (Altarelli et al. 2014). Based on the normalized surface area measures for left and right planum temporale, an asymmetry index (AI) was calculated, AI = (Right−Left)/0.5* (Right + Left). Positive scores indicate a rightward planum temporale lateralization (right > left) and negative scores indicate a leftward lateralization (left > right).

All labeling was done by the first author (J.V.) who was blind to group membership. After first analyses, all labels were evaluated and adjusted with the second author (I.A.), who was also blind to group membership, so as to reach consensus. Additionally, inter- and intra-rater reliability was examined in 10 left hemispherical labels in adolescents, showing a sufficient intra-class correlation coefficient for intra-rater reliability (= 0.751) and inter-rater reliability (= 0.706).

Statistical Analyses

The exact same statistical analyses were applied to the pre-readers and the adolescents’ cohorts. For all tests, the significance level was determined at the two-tailed level (P < 0.05), except for the tests to investigate the strength of asymmetry indices relative to zero, for which significance level was determined at the one-tailed level (P < 0.05). Demographic and behavioral differences between groups were tested using one-way analysis of variance (ANOVA), except for the variable sex, which was analysed with Chi-square.

First, group differences in mean hemispheric surface area were explored using ANOVA tests. As aforementioned, planum temporale surface area measurements were normalized by dividing the planum temporale surface area by the corresponding mean hemispheric surface area. To investigate the influence of having a family risk for dyslexia on left and right PT surface area, an analysis of covariance (ANCOVA) was run with hemisphere as within subject factor, family risk group (2 levels: FRD⁻ and FRD⁺) as between subject factor and sex, raw handedness scores, and raw non-verbal IQ scores as covariates. Significant effects were further explored with post hoc comparisons.

The degree of asymmetry was computed (AI) for the FRD⁻ and FRD⁺ groups and tested by applying one-sample t-test relative to zero. Significance was tested one-tailed. To investigate the influence of having a family risk for dyslexia on planum temporale asymmetry, an ANCOVA was run with the AI as dependent variable, family risk group (FRD⁻ and FRD⁺) as between subject factor and sex, raw handedness scores, and raw non-verbal IQ scores as covariates.

Finally, all analyses were re-run with a reading ability score as covariate. This score was defined as a word and pseudoword composite score, measured in grade 2 for the pre-readers and in grade 9 for the adolescents.

Results

Demographic and Behavioral Results

Tables 1 and 2 show demographic and behavioral data of the FRD⁻ and FRD⁺ participants, after exclusion of all left-handers (n = 54 pre-readers, n = 28 adolescents). There were no significant differences between groups in sex, age, handedness, and non-verbal IQ. One pre-reader and 3 adolescents were ambidextrous. Longitudinal reading data of both cohorts show significant differences between groups for word reading and pseudoword reading.

Planum Temporale in Pre-readers

Effect of Family Risk for Dyslexia

To investigate the effect of having a family risk for dyslexia on planum temporale surface area before the onset of reading acquisition, pre-readers with a family risk for dyslexia (n = 27) were compared with pre-readers without a family risk (n = 27). There were no significant differences between FRD⁻ and FRD⁺ children in mean hemispheric surface area for the left or for the right hemisphere (respectively, F_(1,52) = 0.637, P = 0.428; F_(1,52) = 0.420, P = 0.520). The ANCOVA showed that the effect of hemisphere on planum temporale surface area did not reach significance (F_(1,49) = 3.671, P = 0.061). There was also no effect of family risk group (F_(1,49) = 2.045, P = 0.159), or of the covariates sex (F_(1,49) = 0.026, P = 0.872), handedness (F_(1,49) = 0.519, P = 0.475), and non-verbal IQ (F_(1,49) = 1.655, P = 0.204). There was, however, a significant hemisphere * family risk group interaction (F_(1,49) = 5.182, P = 0.027). Pairwise post hoc comparisons revealed that the interaction was driven by a leftward lateralization in the FRD⁻ group (left: M = 0.0035, SD = 0.0011; right: M = 0.0027, SD = 0.0013; t₍₂₆₎ = 2.720; P = 0.011), while the FRD⁺ group showed a symmetrical distribution of the planum temporale (left: M = 0.0027, SD = 0.0011; right: M = 0.0027, SD = 0.0011; t₍₂₆₎ = 0.012, P = 0.990) (see Fig. 2). Post hoc comparisons also revealed that differences between the FRD⁻ and the FRD⁺ groups were driven by the left (F_(1,52) = 7.187, P = 0.010), but not by the right planum temporale (F_(1,52) = 0.001, P = 0.974).

To investigate the robustness of these results, the analysis was re-run within the NR (n = 42; 25 FRD⁻NR, 17 FRD⁺NR), thereby excluding the effect of reading disability. Interestingly, this analysis confirms the hemisphere * family risk group interaction (F_(1,37) = 9.664, P = 0.004), in the absence of an effect of hemisphere (F_(1,37) = 1.773, P = 0.191), sex (F_(1,37) = 0.659, P = 0.422), handedness (F_(1,37) = 0.551, P = 0.462), or non-verbal IQ (F_(1,37) = 1.621, P = 0.211). Similarly, pairwise post hoc comparisons revealed leftward lateralization in the FRD⁻NR group (left: M = 0.0036, SD = 0.0011; right: M = 0.0026, SD = 0.0013; t₍₂₄₎ = 3.027; P = 0.006), while the FRD⁺NR group showed a symmetrical distribution of the planum temporale (left: M = 0.0026, SD = 0.0012; right: M = 0.0029, SD = 0.0011; t₍₁₆₎ = −1.218, P = 0.241). Differences between the NR groups were again driven by the left (F_(1,40) = 8.343, P = 0.006), but not by the right planum temporale (F_(1,40) = 0.344, P = 0.561). Hence, the effect of family risk on planum temporale surface area is robust and independent of later reading skill, showing a larger proportion of left lateralized cases in children without a familial risk for dyslexia. Controlling for later reading skill in both ANCOVA models, by a word and pseudoword reading composite score in grade 2, did not change any of the results (see also Supplementary information).

Planum Temporale Lateralization

The significance of planum temporale lateralization in all participants was investigated by a one-sample t-test against zero on the asymmetry indices. The mean AI of the FRD⁻ group (M = −0.31, SE = 0.10) was significantly left lateralized (t₍₂₆₎ = −3.133, P = 0.002) (Fig. 2). The planum temporale was not significantly lateralized in the FRD⁺ group (M = 0.03, SE = 0.09, t₍₂₆₎ = 0.378, P = 0.354). In line with the previous results, there was a significant effect of family risk group on planum temporale AI (F_(1,49) = 7.313, P = 0.009), tested with an ANCOVA, while there was no effect of the covariates sex (F_(1,49) = 0.016, P = 0.899), handedness (F_(1,49) = 1.177, P = 0.283), and non-verbal IQ (F_(1,49) = 1.756, P = 0.191). Inter-individual variation in the lateralization index is presented in Figure 3 (top panel). Controlling for later reading skill did not change the results.

Planum Temporale in Adolescents

Effect of Family Risk for Dyslexia

The robustness of the effect of family risk for dyslexia found in pre-readers, who did not have substantial reading experience, was tested in adolescents. There were no significant differences between FRD⁻ (n = 14) and FRD⁺ (n = 14) in mean hemispheric surface area for the left nor right hemisphere (respectively, F_(1,26) = 0.135, P = 0.716; F_(1,26) = 0.224, P = 0.640). For planum temporale surface area, there was no significant effect of hemisphere (F_(1,23) = 2.223, P = 0.150) or family risk group (F_(1,23) = 0.006, P = 0.940), neither was there an effect of the covariates sex (F_(1,23) = 0.075, P = 0.786), handedness (F_(1,23) = 0.112, P = 0.741), and non-verbal IQ (F_(1,23) = 0.146, P = 0.706). Interestingly, similar to the results in pre-readers there was a significant hemisphere * family risk group interaction (F_(1,23) = 6.876, P = 0.015) (The robustness of the results was, similarly to the pre-readers, confirmed when restricting the analysis to the NR participants. Note that in this analysis the FRD⁺NR group is small (n = 4).). Pairwise post hoc comparisons revealed that the interaction was driven by reversed asymmetry patterns in the 2 groups, though each was not significant (Fig. 2). While the FRD⁻ group showed a trend towards a larger left than right planum temporale (left: M = 0.0035, SD = 0.0014; right: M = 0.0028, SD = 0.0016; t₍₁₃₎ = 1.457; P = 0.169), the FRD⁺ group showed a trend towards a larger right than left planum temporale (left: M = 0.0027, SD = 0.0010; right: M = 0.0037, SD = 0.0014; t₍₁₃₎ = −1.934, P = 0.075). Differences between the FRD⁻ and the FRD⁺ groups were not driven specifically by the left (F_(1,26) = 2.997, P = 0.095) or right planum temporale (F_(1,26) = 2.366, P = 0.136), but rather by a combination of an enlarged right planum temporale and a smaller left planum temporale in the FRD⁺ group. Controlling for reading skill in the ANCOVA model, quantified by a word and pseudoword reading composite score in grade 9, did not change the results (see also Supplementary information).

Planum Temporale Lateralization

To investigate the significance of planum temporale lateralization, a one-sample t-test against zero was run on the asymmetry indices. For the FRD⁻ group the mean AI (M = −0.31, SE = 0.21) tended to be left lateralized (t₍₁₃₎ = −1.501, P = 0.079). Similar, the rightward lateralization in the FRD⁺ group (M = 0.26, SE = 0.16) tended to be right lateralized (t₍₁₃₎ = 1.613, P = 0.065) (Fig. 2). Individual differences in the lateralization index are presented in Figure 3 (bottom panel). There was again a significant effect of family risk group on planum temporale AI (F_(1,23) = 5.536, P = 0.028), tested with an ANCOVA, while there was no effect of the covariates sex (F_(1,23) = 0.072, P = 0.791), handedness (F_(1,23) = 0.846, P = 0.367), and non-verbal IQ (F_(1,23) = 1.681, P = 0.208). Controlling for reading skill did not change the results.

Discussion

To the best of our knowledge, this is the first study to examine the influence of having a family risk for dyslexia on planum temporale lateralization, both prior to reading onset and after years of reading experience. Our results reveal that atypical planum temporale asymmetry is related to family history of dyslexia. The analyses uncovered a hemisphere by family risk group interaction on planum temporale surface area and an effect of family risk on planum temporale asymmetry indices. While among individuals without a family risk a greater proportion of leftward lateralization is found, individuals with a family risk for dyslexia show the reversed pattern, that is, a greater proportion of rightward cases. These differences are not driven by reading (dis)ability but rather are specific to familial risk.

An atypical lateralization pattern gives rise to the question of which anatomical features might be driving it. Galaburda et al. (1987) suggested that a reduced leftward asymmetry in individuals with dyslexia might be due to an enlarged right planum. This hypothesis was not fully supported by subsequent MRI studies (Larsen et al. 1990; Altarelli et al. 2014). Our results provide no support for an enlarged right planum temporale in our FRD⁺ groups. Prior to reading onset, the difference between the FRD⁺ and FRD⁻ group seems in fact to be driven by the left planum temporale.

The majority of previous MRI studies on planum temporale anatomical asymmetry in developmental dyslexia did not report on the absence/presence of a family risk for dyslexia in their participants (e.g., Larsen et al. 1990; Hugdahl et al. 1998; Shapleske et al. 1999; Eckert and Leonard 2000; Sanchez-Bloom et al. 2013; Altarelli et al. 2014). A few studies considered having a family risk as an inclusion criterion for the dyslexic group (Hynd et al. 1990; Semrud-Clikeman et al. 1991; Robichon et al. 2000; Eckert et al. 2003), but provided no proper control subjects without a family risk. Hence, besides a confounding effect of handedness, IQ, sex and the applied labeling method, the results of previous reports might have been confounded by the presence/absence of a family risk for dyslexia in the participants. In our study, a highly detailed individual labeling method (Altarelli et al. 2014) was applied and after excluding the potential confounding effect of handedness, sex, non-verbal IQ, and hemispheric surface area, the results show that deviant planum temporale asymmetry is related to family risk for dyslexia.

To achieve a better understanding of the causal pathways leading to dyslexia, it is important to clarify the potential role of planum temporale surface area asymmetry. Bishop (2013) described 4 causal models depicting the hypothetical relation between genes, atypical cerebral lateralization for language, and language and literacy impairments. In our study, family history of dyslexia may be interpreted as a (indirect) proxy for genotype-related properties. Our results, pointing towards the influence of familial risk on planum temporale asymmetry, are in line with the hypothesis that atypical planum temporale asymmetry is not exclusively related to developmental dyslexia (Steinmetz and Galaburda 1991). Planum temporale lateralization alone cannot explain the emergence of dyslexia. Our results suggest that a model in which the effect of susceptibility genes on language is entirely mediated by planum temporale lateralization, as proposed in the “endophenotype model” of Bishop (2013), is unlikely. In addition, results of genetic studies to date have not been conclusive in this regard (for an overview see Bishop 2013) and further research is required. Another option, the “neuroplasticity model”, posits that reading ability itself influences cerebral lateralization (Bishop 2013). The results of our study, showing no relation between planum temporale lateralization and (later) reading ability in pre-readers and adolescents, and showing a difference in lateralization prior to reading onset depending on the family risk status, do not point in that direction either. A third alternative possibility is that atypical planum asymmetry does not play a causal role in the emergence of dyslexia, but planum temporale asymmetry and developmental dyslexia might be both influenced by the same genetic factors (Bishop 2013). Our results do not discard this “pleiotropy model”, as far as planum temporale lateralization is concerned. A relation between family risk and planum temporale asymmetry is shown, even prior to reading onset, as well as a link between family risk and literacy problems, as a large portion of the at-risk participants actually developed dyslexia (37% of the pre-readers and 71% of the adolescents). Having a family risk for dyslexia is related to atypical planum asymmetry and having a family risk increases the chances to develop dyslexia, but no relation necessarily exists between having a reading impairment and presenting an atypical planum temporale lateralization pattern. Finally, a last plausible model theorizes that atypical lateralization may interact with other risk factors for dyslexia (Bishop 2013). Atypical lateralization would thus be one of a number of risk factors for literacy impairment. Some of these may be genetic, others environmental. In light of the results of this study, pointing towards an influence of family risk that is present prior to reading onset and that remains present after years of reading experience, and considering prenatal influences as hypothesized by Galaburda (1993), it appears likely that developmental dyslexia is caused by a complex interplay between genetic and environmental risk factors.

It should be noted that elements in all of the models described above are subject to other sources of variation. What we discuss here is only a simplified version of the seemingly complex relationship between genes, brain lateralization, and dyslexia. Further studies will be needed in order to clarify the role of all elements included in these models. In fact, some studies have already contributed to a partial understanding of the complex relationships between genetics, neurological features, and literacy impairment. Leonard et al. (2001, 2011) work on the neural markers of dyslexia in children and adults suggests that atypical lateralization is a risk factor for literacy impairment, though the influence of genetic factors was not measured in these studies. Evidence for a relation between cerebral organization and phonological skills, dependent on environmental factors, has also been provided (Eckert et al. 2001). Moreover, studies in pre-readers with a family risk for dyslexia have shown structural anomalies in temporoparietal regions (Yamada et al. 2011; Black et al. 2012; Hosseini et al. 2013) and occipitotemporal regions (Specht et al. 2009; Raschle et al. 2011; Raschle et al. 2012), as well as in reading-related white matter pathways (Vandermosten et al. 2015). Though these studies do not specifically investigate brain asymmetries, they are suggestive of a relation between genetic factors and neurological anomalies. The role of these anomalies in the emergence of reading ability, however, and their potential causal influence remain to be explored.

Investigations of the neural markers of developmental dyslexia in relation to reading ability and remediation have attempted to pinpoint the influence of reading status on gray matter volume (Krafnick et al. 2014) and on cortical thickness lateralization (Ma et al. 2015), in children aged 9 and above. These studies found evidence that some of the structural anomalies in key regions of the reading network were independent of reading ability. Recently, an ERP study in children aged 11–12 years has provided evidence that basic auditory processing deficits at the brain level are related to familial risk for dyslexia, rather than to reading disability (Hakvoort et al. 2015). Though the picture is still incomplete, emerging evidence from this body of work suggests that some cerebral anomalies might be present early in development and might be related to familial risk, irrespective of later reading outcome.

It is important to note that in this study, we were only able to replicate previous findings, indicating that 60–70% of the general population shows a leftward asymmetry of the planum temporale (Geschwind and Levitsky 1968; Witelson and Pallie 1973; Wada et al. 1975; Galaburda et al. 1987; Steinmetz and Galaburda 1991; Chance et al. 2006), in the pre-readers and not in the adolescents. In the pre-readers, leftward lateralization was found in the children without a family risk, regardless of their reading status. In adolescents, most of the participants had a larger left than right planum temporale, but the lateralization did not reach significance (P = 0.079). This might be explained by the lateralization of one participant, that could be considered as an outlier (>1.65 SD from the group mean), presenting extreme rightward lateralization (see Fig. 3, bottom panel). Excluding this participant from the analyses results in a significant leftward planum temporale lateralization in the group of adolescents without a family risk for dyslexia (P = 0.022).

In sum, the results of this study suggest that atypical planum temporale asymmetry is related to having a family risk for dyslexia; they show that it is present before the onset of reading acquisition and confirm that it remains present after years of reading experience.

Supplementary Material

Supplementary data is available at Cerebral Cortex online.

Funding

This research was funded by the Research Council of KU Leuven (OT/12/044) and the Research Foundation Flanders (G0920.12).

Notes

Maaike Vandermosten is postdoctoral fellow of the Research Foundation Flanders. François Leroy is gratefully acknowledged for support on data analyses. We would also like to thank Ron Peeters for technical support. We are grateful to all the adolescents and parents who participated in this study. Conflict of Interest: None declared.

References