Original article 9
GRIN2B mediates susceptibility to intelligence quotient and cognitive impairments in developmental dyslexia Sara Mascherettia, Andrea Facoettia,c, Roberto Giordab, Silvana Berib, Valentina Rivaa, Vittoria Trezzia, Maria R. Cellinod and Cecilia Marinoa,e,f Objective(s) Developmental dyslexia (DD) is a complex heritable condition associated with impairments in multiple neurocognitive domains. Substantial heritability has been reported for DD and related phenotypes, and candidate genes have been identified. Recently, a candidate gene for human cognitive processes, that is, GRIN2B, has been found to be associated significantly with working memory in a German DD sample. In this study, we explored the contribution of six GRIN2B markers to DD and key DDrelated phenotypes by association analyses in a sample of Italian nuclear families. Moreover, we assessed potential gene-by-environment interactions on DD-related phenotypes. Materials and methods We carried out a family-based association study to determine whether the GRIN2B gene influences both DD as a categorical trait and its related cognitive traits in a large cohort of 466 Italian nuclear families ascertained through a proband affected by DD. Moreover, we tested the role of the selected GRIN2B markers and a set of commonly described environmental moderators using a test for G × E interaction in sib pairbased association analysis of quantitative traits in 178 Italian nuclear families.
term memory. No significant gene-by-environment effects were found. Conclusion Our results add further evidence in support of GRIN2B contributing toward DD and deficits in DD. More specifically, our data support the view that GRIN2B influences DD as a categorical trait and its related quantitative phenotypes, thus shedding further light on the etiologic basis and the phenotypic complexity of this disorder. Psychiatr Genet 25:9–20 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved. Psychiatric Genetics 2015, 25:9–20 Keywords: association study, developmental dyslexia, developmental dyslexia-related neuropsychological traits, gene-by-environment interaction, GRIN2B, N-methyl-D-aspartate receptors a Child Psychopathology Unit, Department of Child Psychiatry, bMolecular Biology Laboratory, Scientific Institute ‘Eugenio Medea’, Bosisio Parini (Lecco), c Developmental and Cognitive Neuroscience Lab, Department of General Psychology, University of Padua, Padua, dRegional Reference Center for the Specific Learning Disability, ULSS 20, Verona, Italy, eCentre de recherche de l’Institut universitaire en santé mentale de Québec and fDepartment of Psychiatry and Neuroscience, Laval University, Québec, Canada
Correspondence to Sara Mascheretti, PhD, Child Psychopathology Unit, Department of Child Psychiatry, Scientific Institute ‘Eugenio Medea’ via don Luigi Monza, 20 23842 Bosisio Parini (Lecco), Italy Tel: + 39 031 877 813; fax: + 39 031 877 499; e-mail: [email protected]
Results Evidence for a significant association was found with the categorical diagnosis of DD, performance intelligence quotient, phonemic elision, and auditory short-
Received 16 September 2013 Revised 17 May 2014 Accepted 24 October 2014
Introduction
Following earlier descriptions (Hallgren, 1950) of high familial aggregation of the disorder, substantial heritability has been reported for DD (Fisher and De Fries, 2002). Twin studies show a wide variation in heritability estimates (range: 0.18–0.72; Plomin and Kovas, 2005) across DD and DD-related phenotypes, whereas shared environmental influences appear to play a less important role in explaining the familial aggregation of the disorder (Pennington, 1995). The etiology of DD involves multiple interacting risk factors, which can either be genetic or environmental, and underlies a continuously distributed liability. Since the early 1980s, at least nine DD risk loci have been mapped to chromosomes 1, 2, 3, 6, 15, 18, and X (Scerri and Schulte-Körne, 2010), and DYX1C1, KIAA0319, DCDC2, and ROBO1 have been suggested as DD susceptibility genes (Taipale et al., 2003; Francks et al., 2004; Hannula-Jouppi et al., 2005; Meng et al., 2005), although negative findings have also been reported.
Developmental dyslexia (DD) is a complex heritable condition typically diagnosed in the first school years characterized by an impaired reading acquisition despite normal intelligence and adequate educational opportunities (American Psychiatric Association, 1994), and associated with impairments in multiple neurocognitive domains (Gabrieli, 2009). Several data show that the intelligence quotient (IQ) is associated strongly with reading development and impairment (Stanovich, 1986; Shaywitz et al., 1995; Berninger et al., 2001; Ferrer and McArdle, 2004; Ferrer et al., 2007, 2010), although the exact nature of this relationship still remains unclear (Newman et al., 1991; Rispens et al., 1991; Jiménez Glez and Rodrigo López, 1994; Stuebing et al., 2002). Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's website (www.psychgenetics.com). 0955-8829 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
DOI: 10.1097/YPG.0000000000000068
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
10 Psychiatric Genetics 2015, Vol 25 No 1
Recently, a candidate gene for human cognitive processes, that is, GRIN2B, located on chromosome 12p13.1, has been found to be associated significantly with working memory in a German DD sample (Ludwig et al., 2009). In this study, four markers within intron 3 were associated with short-term memory in DD by suggesting that variation within this gene may contribute toward the genetic background of specific DD-related neuropsychological phenotypes (Ludwig et al., 2009). More specifically, the GRIN2B gene codes for a specific subunit composition of N-methyl-D-aspartate (NMDA) receptors – a class of ionotropic glutamate receptors (i.e. GluN2B subunit). Over the past few years, the involvement of N-methyl-D-aspartate receptors (NMDARs) in learning and memory formation has been emphasized (Cull-Candy et al., 2001; Lau and Zukin, 2007; Kalia et al., 2008). NMDAR are ionotropic, glutamatergic receptors involved in excitatory synaptic transmission in the central nervous system (Cull-Candy et al., 2001). In particular, the NMDAR–GluN2B subunit plays a critical role in experience-dependent synaptic plasticity associated with learning and memory (Kutsuwada et al., 1996; Ito et al., 1997; Tang et al., 1999; Kim et al., 2005; Akashi et al., 2009; Fetterolf and Foster, 2011). Animal studies show that the Glun2b subunit is required for neuronal pattern formation in general, and for channel function and formation of dendritic spines in hippocampal pyramidal cells in particular (Ito et al., 1997; Cull-Candy et al., 2001; Kim et al., 2005; Akashi et al., 2009). Transgenic overexpression of Grin2b in the forebrain of mice, and in the cortex and hippocampus of rats results in an increased activation of the NMDARs, with mice and rats showing a superior performance in various tests of learning and memory (Tang et al., 1999; Wang et al., 2009). These findings point toward GRIN2B as a susceptibility gene for complex traits/disorders in humans. Nevertheless, although genetic contributions are clearly relevant to the etiology of these phenotypic expressions, it is widely recognized that additional factors – whose nature cannot be immediately identified as genetic – also act as powerful predictors for human cognition and neurodevelopmental disorders, such as DD. A large range of factors, that is pre/ perinatal (Gilger et al., 1992; Fried et al., 1997; Bowen et al., 2002; Batstra et al., 2003; Samuelsson et al., 2006; Michaelsen et al., 2009; van Baar et al., 2009; Hoque et al., 2012), sociodemographic, and familial (Fergusson and Lynskey, 1993; Fergusson and Woodward, 1999; Melekian, 2001; Hoff and Tian, 2005), seems to impinge on human cognition and DD (Davis et al., 2001; Gayán and Olson, 2001, 2003; Byrne et al., 2002; Petrill et al., 2006; Harlaar et al., 2007), and may contribute toward increasing the likelihood of developing this disorder. Given the above evidence, it thus seems appropriate to complement genomic investigation of such complex traits by taking into account the gene-by-environment interaction (G × E), which is a specific form of interplay, whereby genetic susceptibility conferred by a specified allele is modulated by a measured environmental factor (Rutter et al., 2006).
Here, we sought to resolve which aspects of the phenotypic profile of DD were unambiguously attributable to GRIN2B effects by exploring the contribution of four SNPs that have been shown previously to contribute toward the genetic background of DD (i.e. rs1012586G/C; rs2268119A/T; rs2216128T/C; rs2192973C/T; Ludwig et al., 2009) and two markers (i.e. rs5796555 − /A and rs11609779C/T) close enough to markers rs1012586G/C and rs2216128T/C, respectively, to be amplified and sequenced with them, to DD, defined as a categorical trait, and key DD-related phenotypes by association analyses in a sizeable sample of Italian nuclear families. Finally, we tested the role of the selected GRIN2B markers and a set of commonly described environmental moderators using a test for G × E interaction in a sib pairbased association analysis of quantitative traits (van der Sluis et al., 2008; Mascheretti et al., 2013a).
Materials and methods The protocol was approved by the Scientific Institute ‘Eugenio Medea’ Ethics and Scientific Board. Sample
This study is part of an ongoing project on the genetics of reading disabilities at the Department of Child Psychiatry and Rehabilitation Centre at the Scientific Institute ‘Eugenio Medea’, Bosisio Parini, Italy, and at the Centro Regionale di Riferimento per i Disturbi dell’Apprendimento – CRRDA (‘Regional Reference Center for the Specific Learning Disability’, ULSS 20, Verona, Italy) (Mascheretti et al., 2013a, 2013b, 2014). To date, 466 unrelated nuclear families of probands with DD (738 offspring) have been recruited; except for 47 families that had only one parent available, all parents were represented, yielding a total sample of 1623 individuals, all of Italian ancestry. The ascertainment scheme has been reported in detail elsewhere (Marino et al., 2003). Briefly, nuclear families were recruited if probands met the criteria for DD according to the DSM-IV (American Psychiatric Association, 1994). After parental informed consent, offspring underwent an extensive medical assessment and a battery of tests, which evaluate text, word, nonword reading (Cornoldi, 1995; Sartori et al., 1995; Cornoldi et al., 1998), writing-under-dictation of word, nonword and sentences-containing-homophones (Sartori et al., 1995), forward/backward digit spans (Reynolds and Bigler, 1994), phonemic elision and blending (Cossu et al., 1988), mathematics abilities (Cornoldi et al., 2003; Cornoldi and Lucangeli, 2004), and the Wechsler Intelligence Scale for Children, Revised (WISC-R) (Wechsler, 1981) or the Wechsler Intelligence Scale for Children, 3rd ed. (WISC-III) (Wechsler, 2006). For all tests, standardized scores on the Italian population are provided (for an extensive description of tests, see Supplemental digital content 1, http://links.lww.com/PG/
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
GRIN2B and susceptibility to dyslexia Mascheretti et al. 11
A119). Sibs were administered only two subtests of the intelligence scale, that is vocabulary and block design, which shows a high correlation (r) with, respectively, verbal IQ (r = 0.82; Wechsler, 1981, 2006) and performance IQ (r = 0.73; Wechsler, 1981, 2006). The criteria used to define the probands’ affection status were as follows: (a) a performance on a timed text-reading test at least 2 SD below the expected grade level mean on either accuracy or speed; or (b) an absolute score at least 2 SD below the expected grade level mean on accuracy or speed in a reading list of unrelated words or nonwords; and (c) IQ more than 84. Siblings were included if they were fully biological, older than 6 years and younger than 18 years, if they had no history of neurological and sensorial disorders, and if the mean score of vocabulary and block design subtests was above 7, irrespective of their reading performance. Blood or mouthwash samples were obtained from all offspring and their biological parents.
Phenotypes
We considered eight phenotypes in the single-marker transmission disequilibrium test (TDT) analyses: (1) DD as a discrete trait. (2) READING, as measured by averaging speed (s) and accuracy (expressed in number of errors) gradestandardized scores in text, word, and nonword reading tasks (Cornoldi, 1995; Sartori et al., 1995; Cornoldi et al., 1998) as the mean bivariate correlation among the above-mentioned tasks was considerable (r = 0.450). (3) SPELLING, as measured by averaging the accuracy (expressed in number of errors), grade-standardized scores in writing-under-dictation word, nonword, and sentences-containing-homophones (bivariate correlation among the above-mentioned tests, r = 0.563; Sartori et al., 1995). (4) Performance IQ, as measured by the WISC-R or the WISC-III. For siblings, the ‘block design’ subtest was used to provide a prorated, performance IQ score (Wechsler, 1981, 2006). (5) Auditory STM, as measured by averaging agenormed scores in the forward and backward digit spans (bivariate correlation among the abovementioned spans, r = 0.701; Reynolds and Bigler, 1994). (6) Phonemic elision (ELISION; Cossu et al., 1988). (7) Phonemic blending (BLENDING; Cossu et al., 1988). (8) Mathematical abilities (Cornoldi et al., 2003; Cornoldi and Lucangeli, 2004), as measured by a principalcomponent-analysis factor that accounted for 53% of the total test variance among mental and written calculation, number dictation, and numerical facts (mean factor loadings = 0.64 ± 0.08; MATHEMATICS).
Of the total sample (n = 466; see the Sample section), (i) 324 families (538 offspring) had complete measures of READING, SPELLING, performance IQ, auditory STM (group 1), and (ii) 174 families (223 offspring) had complete measures of ELISION, BLENDING and MATHEMATICS (group 2). Group 1 and group 2 overlapped with respect to the phenotypic data available.
Environmental data collection
Selection of the environmental factors has been described extensively elsewhere (Mascheretti et al., 2013a, 2013b). Briefly, using a large case–control sample, we removed all variables with an at-risk category frequency under 5% and collapsed those variables that were highly correlated to obtain more concise and comprehensive variables (Mascheretti et al., 2013a, 2013b). Environmental factors included in further analyses were as follows: (1) Maternal smoking – hereafter smoke. (2) Risk of miscarriage during pregnancy – hereafter miscarriage. (3) Birth weight. (4) Breast feeding. (5) Parental age, as an average of the father and the mother’s age at child’s birth. (6) Socioeconomic status, defined by parental employment. (7) Parental education, as an average of the father and the mother’s educational qualifications. Dichotomous responses (yes/no) were coded with ‘0’ when answers were ‘no’ and ‘1’ when answers were ‘yes’, except for breast feeding, which was coded as ‘0’ if child had been breastfed. The employment response was coded according to the Hollingshead nine-point scale (Hollingshead, 1975). A score (from 10 to 90) was assigned to each job; the higher of two scores was used when both parents were employed. The educational qualification response was scored according to a ninepoint ordinal scale on the basis of the Italian school system (range between 10, corresponding to fifth-grade elementary school, and 90, equivalent to a postdoctoral degree). One-hundred and seventy-eight families (144 pairs, 31 triplets, two with four siblings, and one with five siblings, 394 offspring) had at least one sibling and complete environmental data (group 3); in this latter group, complete phenotypes were available for READING, SPELLING, performance IQ, and auditory STM, whereas only 60 families had complete data for ELISION, BLENDING, and MATHEMATICS, which were therefore excluded from G × E analyses because of limited power. As for group 2, group 1 and group 3 overlapped with respect to the phenotypic data available. Descriptive statistics of the environmental variables are outlined in Supplemental digital content 2 (http://links. lww.com/PG/A120). All environmental factors, except
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
12 Psychiatric Genetics 2015, Vol 25 No 1
SMOKE, MISCARRIAGE, breast feeding, and parental marital status, were z-transformed on the basis of the sample norms to avoid a scaling effect in further analyses.
Fig. 1
Genotype error checking was completed in PEDSTATS (Wigginton and Abecasis, 2005) and inconsistent genotypes were zeroed out and were not considered for further analysis. Families were excluded in the case of multiple inconsistent genotypes. Allelic frequencies and Hardy– Weinberg equilibrium for the markers under consideration were calculated in parents (Supplemental digital content 3, http://links.lww.com/PG/A121). No significant deviations from allele frequencies reported in Ludwig et al. (2009) were found (Supplemental digital content 3, http://links. lww.com/PG/A121). For all the six genotyped SNPs, P-values for deviation from the Hardy–Weinberg equilibrium were not lower than 0.008 (P = 0.05/6; Supplemental digital content 3, http://links.lww.com/PG/A121), and thus no SNPs were excluded. The linkage disequilibrium structure of GRIN2B was analyzed using only the parental genotypes; linkage disequilibrium was extracted and plotted in Haploview 4.0 (Fig. 1). As shown in Fig. 1, even if markers rs5796555 − A and rs11609779C/T were typed as a byproduct of rs1012586G/C and rs2216128T/C, respectively, they were not excluded from further analysis as they did not show correlation with any of other genotyped SNPs. Statistical analysis Genetic association analysis
Genetic association was investigated using a family-based association test, that is, the quantitative transmission disequilibrium test (QTDT, version 2.5.1) as modeled by Abecasis et al. (2000), which allows to control for
3
39
20 8
Block 1 (13 kb) 4 5
3
6 5
6
25
3
rs2192973
2
rs11609779
rs2268119
1
rs2216128
rs1012586
Amplification and sequencing of four portions of the human GRIN2B gene allowed typing of the following SNPs: rs5796555 − /A; rs1012586G/C; rs2268119A/T; rs2216128T/C; rs11609779C/T; rs2192973C/T. In particular, we chose to genotype the four SNPs that in a previous study were found to be associated significantly with short-term memory in a German DD sample (i.e. rs1012586, rs2268119, rs2216128, and rs2192973; Ludwig et al., 2009), suggesting that variations in GRIN2B may contribute toward the genetic background of specific cognitive processes that are correlated to DD. Markers rs5796555 − /A and rs11609779C/T are close enough to markers rs1012586G/C and rs2216128T/C, respectively, to be amplified and sequenced with them. Amplifications were performed in 10-μl reactions using JumpStart Red ACCUTaq LA DNA polymerase (Sigma, St. Louis, Missouri 63103, USA) and the following protocol: 30 s at 96°C, 35 cycles of 15 s at 94°C/20 s at 58°C/30 s at 68°C, and a 5 min final elongation time. Sequencing reactions were performed using a Big Dye Terminator Cycle Sequencing kit (Applied Biosystems, St. Louis, Missouri 63103, USA) and run on an ABI Prism 3130xl Genetic Analyzer. (Primers are available on request from the authors.)
rs5796555
Genotyping
91
3 26
1 4
1 4
Haploview plot showing pairwise linkage disequilibrium (r2 values) for six SNPs of GRIN2B on the basis of parents’ genotypes. Numbers in squares are r2 values.
population stratification bias and for the analyses of quantitative traits. Single-marker TDT analyses were carried out for DD as a discrete trait using the ‘-ad’ option, which allows to test using only affected individuals. Quantitative traits were analyzed using the ‘-wega’ option, which allows to adopt commonly used variance components, that is, the environmental variance (e), polygenic variance (g), and additive major locus (a). Genetic findings are not presented for the entire set of families because of missing data and because TDTs require individuals to have heterozygous ancestors in the pedigree; thus, offspring of homozygous parents are uninformative (Abecasis et al., 2000). We calculated the genetic association for the total sample. Moreover, as follow-up association signals over different severity groups may confirm a possible true contribution of the selected SNPs to DD and DD-related performance (Ludwig et al., 2009), we tested the genetic association for a subsample selected by severity, that is, by selecting only the nuclear families in which at least one offspring scored less than or equal to 2.50 SD below the general population mean on either accuracy or speed in either text-reading, or word-reading, or nonword reading tasks (n = 292). Tables 1 and 2 show the descriptive statistics of all traits in groups 1 and 2 of the total sample and in the selected-by-severity subsample, respectively. Only empirical P-values are reported, which are computed from 10 000 Monte-Carlo permutations by entering all
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Measures are in SD units, relative to the age-appropriate Italian population norm. Auditory STM, the average between forward/backward digit spans; MATHEMATICS, the regressed factor scores obtained by the PCA; READING, the average between text-reading, word-reading, and nonword reading tasks (both accuracy and speed); SPELLING, the average between word, nonword spelling and orthographic choice tasks. a Age was expressed in years. b Percentage of males was reported. c For siblings, performance intelligence quotient (IQ) represents an estimated score obtained from the ‘Block Design’ subtest as described in the text. d These data were collected on 174 nuclear families.
− 0.34 17.29 5.63 1.28 11.92 1.29 2.61 0.09 − 3.20 − 2.21 − 0.04 − 2.98 − 1.14 − 1.42 (17.37) (1.89) (2.10) (1.04) (1.34) (2.57) (0.82) 109.58 − 0.95 − 0.99 − 0.36 − 0.71 − 2.80 0.51 157 1.03 0.88 2.50 0.59 − 0.17 1.55 65 − 15.43 − 11.30 − 3.33 − 7.74 − 12.20 − 2.49 1.22 6.01 4.42 0.39 1.29 − 0.10 0.57 0.59 − 1.93 − 2.00 0.87 − 1.13 − 0.75 − 0.90 (10.86) (1.72) (3.20) (1.15) (1.40) (2.80) (1.00) 105.58 − 2.53 − 2.63 − 0.28 − 1.22 − 3.66 − 0.14 160 0.09 1.29 3.17 0.88 − 0.17 1.51 80 − 11.52 − 16.30 − 3.33 − 6.25 − 11.44 − 3.40 0.28 6.93 5.63 0.79 3.26 0.16 0.71 0.37 − 1.83 − 2.16 0.50 − 1.57 − 0.85 − 0.96 (15.11) (1.95) (2.92) (1.11) (1.39) (2.75) (1.00) 107.38 − 1.90 − 1.96 − 0.32 − 1.06 − 3.40 0.00 160 1.03 1.29 3.17 0.88 − 0.17 1.55 65 − 15.43 − 16.30 − 3.33 − 7.74 − 12.20 − 3.40 Performance IQ READING SPELLING Auditory STM ELISIONd BLENDINGd MATHEMATICSd
Kurtosis Skew. Mean (SD) Maximum Minimum
Maximum
Mean (SD)
Skew.
Kurtosis
Minimum
Siblings Age = 12.55 (± 3.92)a; sex = 51.5%b
Kurtosis Skew. Mean (SD) Maximum
where τi is the family-specific intercept, ab and aw are the estimated between-family and within-family additive genetic effects of the marker, according to the orthogonal decomposition, e represents the effects of the environmental risk factor, ibg and iwg represent the between-family and within-family effects of the interaction of genotype g and the environmental risk factor, and εij is the residual term (van der Sluis et al., 2008). As such, iwg becomes an estimate of the G × E effect inasmuch as it indicates the change in the allelic variant marker association with the phenotype across the different ecological niches, and it represents the variation of SD units of the z-score on the neuropsychological composite. For instance, a positive value represents an increase in the effect of each additional
Minimum
yijg¼ ti þab Abi þaw Awij þeEi þibg Eij Abi þiwg Eij Awij þeij ;
c
The complete description of this statistical model has been reported in detail elsewhere (see Supplemental digital content 4, http://links.lww.com/PG/A122; Mascheretti et al., 2013a). Briefly, this is an extension of the Fulker et al. (1999) maximum likelihood variance components analysis of quantitative traits in sib-pairs data that incorporates environmental main effects plus G × E effects (van der Sluis et al., 2008). The association effect is orthogonally decomposed here into between-family (b) and withinfamily (w) effects, that is, the expectation of each sib genotype conditional on family genotype data and the deviation from this expectation for each offspring, respectively. To model the interaction effect, we used the sibling-specific value for each environmental risk variable; the phenotypic score (i.e. the observed phenotypic score y for participant j from family i with genotype g) is then modeled as follows:
Probands Age = 10.45 (2.77)a; sex = 72.7%b
To explore the combined role of candidate genes markers and measured environmental factors on performance IQ and DD-related neuropsychological skills (i.e. READING, SPELLING, and Auditory STM), we analyzed the effects of G × E through a general test for G × E interaction in sib pair-based association analysis of quantitative traits (van der Sluis et al., 2008). Table 3 shows the descriptive statistics of the selected neuropsychological traits in group 3.
Total sample Age = 11.36 (3.48)a; sex = 64.9%b
Gene-by-environment interaction analysis
Table 1
the selected markers (see the Materials and methods and Genotyping sections) and the above-described phenotypes (see the Materials and methods and Phenotypes sections) at the same time. Bonferroni correction for multiple testing was not applied because it would have been too conservative (Deffenbacher et al., 2004; Francks et al., 2004; Cope et al., 2005; Meng et al., 2005; Schumacher et al., 2006; Brkanac et al., 2007; Marino et al., 2012). Thus far, we decided to adjust the significance levels by the false discovery rate (FDR) method (Storey, 2002) applied to the seven neuropsychological traits analyzed for each marker, separately for each SNP.
Descriptive statistics of the neuropsychological measures in groups 1 and 2 of the total sample (n = 324 and 174, respectively) with complete phenotypic and genetic information
GRIN2B and susceptibility to dyslexia Mascheretti et al. 13
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Measures are in SD units, relative to the age-appropriate Italian population norm. Auditory STM represents the average between single number forward/backward spans; MATHEMATICS, the regressed factor scores obtained by the PCA; READING, the average between text-reading, word-reading and nonword reading tasks (both accuracy and speed); SPELLING, the average between word, nonword spelling and orthographic choice tasks. a Age was expressed in years. b Percentage of males was reported. c For siblings, performance intelligence quotient (IQ) represents an estimated score obtained from the ‘Block Design’ subtest, as described in the text. d These data were collected on 150 nuclear families.
− 0.38 15.59 5.04 1.30 11.11 1.99 2.46 − 0.11 − 3.03 − 2.10 0.03 − 2.91 − 1.35 − 1.41 150 1.03 0.88 2.50 0.59 − 0.17 1.55 65 − 15.43 − 16.30 − 3.33 − 7.74 − 12.20 − 3.40
160 1.03 1.29 3.17 0.59 − 0.17 1.55
107.58 − 2.12 − 2.07 − 0.31 − 1.13 − 3.36 − 0.03
(14.73) (1.99) (3.04) (1.11) (1.44) (2.78) (1.01)
0.25 − 1.76 − 2.09 0.54 − 1.52 − 0.85 − 0.95
0.13 6.65 5.08 0.70 2.99 0.06 0.63
80 − 11.52 − 16.30 − 3.33 − 6.25 − 11.44 − 3.40
160 − 0.11 1.29 3.17 0.59 − 0.17 1.51
105.66 − 2.76 − 2.74 − 0.28 − 1.31 − 3.68 − 0.17
(12.61) (1.71) (3.35) (1.16) (1.42) (2.83) (1.01)
0.58 − 2.00 − 1.91 0.82 − 1.07 − 0.69 − 0.88
1.13 6.20 3.86 0.28 1.14 − 0.33 0.47
65 − 15.43 − 11.30 − 3.33 − 7.74 − 12.20 − 2.49
109.98 − 1.12 − 1.08 − 0.36 − 0.73 − 2.67 0.44
(16.74) (1.98) (2.18) (1.05) (1.40) (2.57) (0.84)
Kurtosis Skew. Mean (SD) Maximum Skew. Mean (SD) Maximum Skew. Mean (SD) Maximum Minimum
Total sample Age = 11.40 (3.49)a; sex = 62.5%b
Kurtosis
Minimum
Probands Age = 10.58 (2.83)a; sex = 70.5%b
Kurtosis
Siblings Age = 12.50 (± 3.96)a; sex = 52.2%b
Minimum
c
Performance IQ READING SPELLING Auditory STM ELISIONd BLENDINGd MATHEMATICSd
Table 2
Descriptive statistics of the neuropsychological measures in the selected-by-severity subsample (n = 292) with complete phenotypic and genetic information
14 Psychiatric Genetics 2015, Vol 25 No 1
transmission of the minor allele on the neuropsychological composite z-score when environmental variable increases. A negative value of iwg represents a decrease of the same effect. The permutation procedure was repeated 1000 times for each analysis. We decided to adjust the significance levels using the FDR method (Storey, 2002) applied to the 28 tests performed for each marker (7 environmental variable × 4 phenotypes), separately for each marker (Mascheretti et al., 2013a). Sex was taken into account in the extended equation because probands’ sex ratio (males : females) in our sample was almost 3 : 1, and it may imply differences in the mean scores between males and females. Moreover, as simulation studies showed specific situations in which dichotomized variables performed as well as or better than the original quantitative factors, we subsequently decided to dichotomize raw scores of quantitative environmental variables (i.e. birth weight, parental age, socioeconomic status, and parental education; see the Materials and methods and the Environmental data collection sections) according to well-defined cut-off points (for more information, see Supplemental digital content 4, http://links.lww.com/PG/A122).
Results Genetic association analyses with developmental dyslexia and developmental dyslexia-related quantitative traits
Significant associations were found between DD as a discrete trait and the minor alleles ‘A’ and ‘G’ of the markers rs5796555 − /A and rs1012586G/C, respectively, in the total sample (informative families = 331, T = − 2.04, SD = 8.63, nominal P-value = 0.043, and informative families = 347, T = − 1.99, SD = 8.65, nominal P-value = 0.047, respectively). In the selected-by-severity subsample DD as a discrete trait was associated significantly with the minor allele ‘A’ of the marker rs5796555 − /A (informative families = 227, T = − 1.99, SD = 6.65, nominal P-value = 0.048), whereas only a trend toward significance was observed for the rs1012586G/C marker (informative families = 234, T = − 1.79, SD = 6.84, nominal P-value = 0.076). Tables 4 and 5 show both the empirical P-values obtained after implementing 10 000 Monte-Carlo permutations and the q-values after FDR correction. Markettrait associations for READING and SPELLING were nonsignificant in both group 1 and in the selected-by-severity subsample (Table 4). Performance IQ was associated significantly with the minor alleles ‘T’, ‘C’, and ‘T’ of markers rs2268119A/T (χ2 = 6.44; nominal P-value = 0.011; empirical P-value = 0.009; q-value = 0.004; 224 informative families; genetic effect = − 4.714; Table 4), rs2216128T/C (χ2 = 12.38; nominal P-value < 0.001; empirical P-value = 0.001; q-value = 0.007; 227 informative families; genetic effect = − 6.178; Table 4), and rs2192973C/T (χ2 = 10.04; nominal P-value = 0.002; empirical P-value = 0.001; q-value = 0.007; 211 informative families; genetic effect = − 3.460: Table 4),
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
GRIN2B and susceptibility to dyslexia Mascheretti et al. 15
Descriptive statistics of the neuropsychological measures in group 3 of the total sample (n = 178) with complete phenotypic and genetic information
Table 3
Probands (n = 178) Siblings (n = 216) Affected (n = 265) Age = 10.52 (2.81)a; Age = 12.43 (3.81)a; Age = 13.17 (3.12)a; sex = 69.1%b sex = 52.8%b sex = 64.2%b c
Total sample (n = 394) Age = 11.56 (3.51)a; sex = 60.5%b Minimum Maximum Performance IQd READING SPELLING MEMORY NUMBERS
44 − 15.43 − 16.30 − 3.33
160 1.03 1.16 3.00
Mean (SD) 108.41 − 1.66 − 1.75 − 0.42
(16.36) (2.00) (2.92) (1.04)
Not-affected (n = 120) Age = 12.84 (3.94)a; sex = 52.5%b
Skew.
Kurtosis
Mean
SD
Mean
SD
Mean
SD
Mean
SD
0.12 − 2.21 − 2.27 0.43
0.58 8.89 6.11 1.39
107.26 − 2.42 − 2.59 − 0.46
14.80 1.77 3.38 1.06
109.21 − 1.01 − 1.04 − 0.39
17.36 1.96 2.25 1.03
106.53 − 2.39 − 2.40 − 0.53
15.09 1.98 3.22 1.00
111.96 − 0.06 − 0.37 − 0.23
18.00 0.71 1.35 1.08
Measures are in SD units, relative to the age-appropriate Italian population norm. MEMORY NUMBERS, the average between single number forward/backward spans; READING, the average between text-reading, word-reading, and nonword reading tasks (both accuracy and speed); SPELLING, the average between word, nonword spelling, and orthographic choice tasks. a Age was expressed in years. b Percentage of males was reported. c The affection status was assigned according to the criteria outlined in the text. d For siblings, performance intelligence quotient (IQ) represents an estimated score obtained from the ‘Block Design’ subtest, as described in the text.
respectively, in group 1. In the selected-by-severity subsample, performance IQ was associated significantly with the minor allele ‘C’ of marker rs2216128T/C (χ2 = 8.31; nominal P-value = 0.004; empirical P-value = 0.004; q-value = 0.024; 140 informative families; genetic effect = − 6.509; Table 4), whereas only a trend toward significance was found for the rs2268119A/T and rs2192973C/T markers (Table 4). Auditory STM yielded a significant association with the minor allele ‘A’ of the rs5796555 − /A marker in group 1 (χ2 = 5.15; nominal P-value = 0.023; empirical P-value = 0.040; q-value = 0.037; 244 informative families; genetic effect = − 0.240; Table 4). ELISION was associated significantly with the minor allele ‘G’ of the rs1012586G/C marker in the selected-by-severity sample (χ2 = 3.37; nominal P-value = 0.068; empirical P-value = 0.037; q-value = 0.041; 118 informative families; genetic effect = − 0.427; Table 5). BLENDING and MATHEMATICS did not yield any significant association with any markers (Table 5) in both group 2 and the selected-by-severity subsample.
Gene-by-environment interaction analysis
We analyzed G × E interaction effects between six markers and seven environmental factors using the general test for G × E interaction upon READING, SPELLING, performance IQ, and auditory STM composites in group 3. Although we found G-E combinations whose empirical P-values of the iwg term was significant, none survived after FDR correction (Supplemental digital content 5, http://links.lww.com/PG/A123). With respect to the investigation of the sex effect, no empirical P-values were significant (Supplemental digital content 5, http://links. lww.com/PG/A123). Finally, we analyzed G × E interaction effects between six markers and the four dichotomized quantitative environmental factors (see the Materials and methods, the ‘Statistical analysis’, the Gene-by-environment interaction analysis sections, and Supplemental digital
content 4, http://links.lww.com/PG/A122) using the general test for G × E interaction upon the same neuropsychological phenotypes (i.e. READING, SPELLING, performance IQ, and auditory STM) in group 3. We found G × E combinations whose empirical P-values of the iwg term were significant and that showed 33% overlap with the results obtained by implementing quantitative variables, but none survived after FDR correction (Supplemental digital content 6, http://links.lww.com/PG/ A124). No significant sex effect was found (Supplemental digital content 6, http://links.lww.com/PG/A124).
Discussion The ability to read relies on a complex, highly integrated, large-scale network of different cognitive processes. Recently, research applied quantitative cognitive measures to dissect the heterogeneous DD phenotype with respect to the genetic contribution (Marlow et al., 2003; Paracchini et al., 2007; Schulte-Körne et al., 2007). This study was primarily designed to replicate previous findings of the involvement of the GRIN2B gene in DD and its related neuropsychological phenotypes in a sizable sample of Italian nuclear families ascertained for DD. We therefore investigated both linear and G × E interplay effects on DD and DD-related quantitative phenotypes. QTDT analyses yielded significant associations between markers spanning within GRIN2B and DD as a categorical trait and its related quantitative phenotypes, that is, performance IQ, auditory STM, and phonemic elision. These findings suggest that GRIN2B can account for not only part of the disabilities proper of DD, defined as a categorical trait, but also for part of the reduced skills in key DD-related cognitive domains that are observable in these children. Indeed, DD is a heterogeneous disorder in which various phenotypic dimensions are involved, including phonological processing and memory processes (Gabrieli, 2009). As such, our
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
262
213
244
206
311
217
324
159
3.41 (341)
5.15 (400)
0.05 (326)
0.95 (499)
0.68 (350)
0.55 (529)
0.02 (247)
0.35 (415)
χ2 (d.f.)
0.040 0.037b 0.078 0.546b
0.395 0.070b 0.790 0.900b
0.461 0.070b 0.434 0.900b
0.497 0.070b 0.900 0.900b
Empirical P-value
229
260
216
326
227
342
158
265
Informative families
2.31 (342)
2.48 (399)
0.50 (327)
0.00 (499)
3.93 (351)
2.41 (529)
0.18 (247)
1.07 (413)
χ2 (d.f.)
rs1012586G/C
0.114 0.132b 0.134 0.050b
0.981 0.386b 0.378 0.070b
0.209 0.132b 0.135 0.050b
0.239 0.132b 0.653 0.104b
Empirical P-value
184
212
167
261
184
282
134
224
Informative families
0.30 (334)
0.21 (388)
0.76 (315)
0.31 (484)
0.96 (339)
0.80 (514)
4.23 (238)
6.44 (401)
χ2 (d.f.)
rs2268119A/T
0.574 0.038b 0.498 0.185b
0.541 0.038b 0.339 0.163b
0.418 0.038b 0.358 0.163b
0.009 0.004b 0.055 0.122b
Empirical P-value
191
221
174
263
187
282
140
227
Informative families
0.65 (333)
0.14 (388)
0.06 (314)
0.09 (484)
0.00 (338)
0.63 (514)
8.31 (236)
12.38 (400)
χ2 (d.f.)
rs2216128T/C
0.776 0.776b 0.567 0.689b
0.763 0.776b 0.810 0.820b
0.424 0.742b 0.964 0.837b
0.001 0.007b 0.004 0.024b
Empirical P-value
164
195
162
246
187
263
128
208
Informative families
0.05 (333)
0.02 (389)
3.22 (315)
2.02 (486)
0.00 (338)
0.02 (516)
002 (237)
0.48 (402)
χ2 (d.f.)
rs11609779C/T
0.862 0.887b 0.790 0.964b
0.121 0.847b 0.024 0.168b
0.887 0.887b 0.964 0.964b
0.487 0.852b 0.881 0.964b
Empirical P-value
173
203
159
244
171
262
128
211
Informative families
0.59 (333)
0.11 (387)
0.01 (314)
0.01 (483)
0.06 (338)
0.11 (513)
6.73 (238)
10.04 (401)
χ2 (d.f.)
rs2192973C/T
0.810 0.941b 0.615 0.796b
0.941 0.941b 0.932 0.862b
0.731 0.941b 0.813 0.862b
0.001 0.007b 0.009 0.058b
Empirical P-value
Auditory STM, the average between forward/backward digit spans; READING, the average between text-reading, word-reading, and nonword reading tasks (both accuracy and speed); SPELLING, the average between word, nonword spelling, and orthographic choice tasks. a For siblings, performance intelligence quotient (IQ) represents an estimated score obtained from the ‘Block Design’ subtest, as described in the text. b FDR correction’s q-values are bold if ≤ 0.05.
Severity sample
Auditory STM Total sample
Severity sample
SPELLING Total sample
Severity sample
READING Total sample
Severity sample
Informative families
rs5796555 − /A
GRIN2B marker
Market-trait association empirical P-values in both group 1 of the total sample (n = 324) and the selected-by-severity subsample (n = 292) for GRIN2B markers
Performance IQa Total sample
Table 4
16 Psychiatric Genetics 2015, Vol 25 No 1
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
0.041 0.144a 0.101 0.327a 3.98 (225)
2.61 (195) 103
124
0.479 0.852a 0.798 0.964a 0.07 (195)
0.53 (227) 3.38 (226)
2.85 (195)
139
118
0.65 (226) 117 0.37 (238)
MATHEMATICS, the regressed factor scores obtained by the PCA. a FDR correction’s q-values are bold if
GRIN2B mediates susceptibility to intelligence quotient and cognitive impairments in developmental dyslexia Sara Mascherettia, Andrea Facoettia,c, Roberto Giordab, Silvana Berib, Valentina Rivaa, Vittoria Trezzia, Maria R. Cellinod and Cecilia Marinoa,e,f Objective(s) Developmental dyslexia (DD) is a complex heritable condition associated with impairments in multiple neurocognitive domains. Substantial heritability has been reported for DD and related phenotypes, and candidate genes have been identified. Recently, a candidate gene for human cognitive processes, that is, GRIN2B, has been found to be associated significantly with working memory in a German DD sample. In this study, we explored the contribution of six GRIN2B markers to DD and key DDrelated phenotypes by association analyses in a sample of Italian nuclear families. Moreover, we assessed potential gene-by-environment interactions on DD-related phenotypes. Materials and methods We carried out a family-based association study to determine whether the GRIN2B gene influences both DD as a categorical trait and its related cognitive traits in a large cohort of 466 Italian nuclear families ascertained through a proband affected by DD. Moreover, we tested the role of the selected GRIN2B markers and a set of commonly described environmental moderators using a test for G × E interaction in sib pairbased association analysis of quantitative traits in 178 Italian nuclear families.
term memory. No significant gene-by-environment effects were found. Conclusion Our results add further evidence in support of GRIN2B contributing toward DD and deficits in DD. More specifically, our data support the view that GRIN2B influences DD as a categorical trait and its related quantitative phenotypes, thus shedding further light on the etiologic basis and the phenotypic complexity of this disorder. Psychiatr Genet 25:9–20 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved. Psychiatric Genetics 2015, 25:9–20 Keywords: association study, developmental dyslexia, developmental dyslexia-related neuropsychological traits, gene-by-environment interaction, GRIN2B, N-methyl-D-aspartate receptors a Child Psychopathology Unit, Department of Child Psychiatry, bMolecular Biology Laboratory, Scientific Institute ‘Eugenio Medea’, Bosisio Parini (Lecco), c Developmental and Cognitive Neuroscience Lab, Department of General Psychology, University of Padua, Padua, dRegional Reference Center for the Specific Learning Disability, ULSS 20, Verona, Italy, eCentre de recherche de l’Institut universitaire en santé mentale de Québec and fDepartment of Psychiatry and Neuroscience, Laval University, Québec, Canada
Correspondence to Sara Mascheretti, PhD, Child Psychopathology Unit, Department of Child Psychiatry, Scientific Institute ‘Eugenio Medea’ via don Luigi Monza, 20 23842 Bosisio Parini (Lecco), Italy Tel: + 39 031 877 813; fax: + 39 031 877 499; e-mail: [email protected]
Results Evidence for a significant association was found with the categorical diagnosis of DD, performance intelligence quotient, phonemic elision, and auditory short-
Received 16 September 2013 Revised 17 May 2014 Accepted 24 October 2014
Introduction
Following earlier descriptions (Hallgren, 1950) of high familial aggregation of the disorder, substantial heritability has been reported for DD (Fisher and De Fries, 2002). Twin studies show a wide variation in heritability estimates (range: 0.18–0.72; Plomin and Kovas, 2005) across DD and DD-related phenotypes, whereas shared environmental influences appear to play a less important role in explaining the familial aggregation of the disorder (Pennington, 1995). The etiology of DD involves multiple interacting risk factors, which can either be genetic or environmental, and underlies a continuously distributed liability. Since the early 1980s, at least nine DD risk loci have been mapped to chromosomes 1, 2, 3, 6, 15, 18, and X (Scerri and Schulte-Körne, 2010), and DYX1C1, KIAA0319, DCDC2, and ROBO1 have been suggested as DD susceptibility genes (Taipale et al., 2003; Francks et al., 2004; Hannula-Jouppi et al., 2005; Meng et al., 2005), although negative findings have also been reported.
Developmental dyslexia (DD) is a complex heritable condition typically diagnosed in the first school years characterized by an impaired reading acquisition despite normal intelligence and adequate educational opportunities (American Psychiatric Association, 1994), and associated with impairments in multiple neurocognitive domains (Gabrieli, 2009). Several data show that the intelligence quotient (IQ) is associated strongly with reading development and impairment (Stanovich, 1986; Shaywitz et al., 1995; Berninger et al., 2001; Ferrer and McArdle, 2004; Ferrer et al., 2007, 2010), although the exact nature of this relationship still remains unclear (Newman et al., 1991; Rispens et al., 1991; Jiménez Glez and Rodrigo López, 1994; Stuebing et al., 2002). Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's website (www.psychgenetics.com). 0955-8829 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
DOI: 10.1097/YPG.0000000000000068
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
10 Psychiatric Genetics 2015, Vol 25 No 1
Recently, a candidate gene for human cognitive processes, that is, GRIN2B, located on chromosome 12p13.1, has been found to be associated significantly with working memory in a German DD sample (Ludwig et al., 2009). In this study, four markers within intron 3 were associated with short-term memory in DD by suggesting that variation within this gene may contribute toward the genetic background of specific DD-related neuropsychological phenotypes (Ludwig et al., 2009). More specifically, the GRIN2B gene codes for a specific subunit composition of N-methyl-D-aspartate (NMDA) receptors – a class of ionotropic glutamate receptors (i.e. GluN2B subunit). Over the past few years, the involvement of N-methyl-D-aspartate receptors (NMDARs) in learning and memory formation has been emphasized (Cull-Candy et al., 2001; Lau and Zukin, 2007; Kalia et al., 2008). NMDAR are ionotropic, glutamatergic receptors involved in excitatory synaptic transmission in the central nervous system (Cull-Candy et al., 2001). In particular, the NMDAR–GluN2B subunit plays a critical role in experience-dependent synaptic plasticity associated with learning and memory (Kutsuwada et al., 1996; Ito et al., 1997; Tang et al., 1999; Kim et al., 2005; Akashi et al., 2009; Fetterolf and Foster, 2011). Animal studies show that the Glun2b subunit is required for neuronal pattern formation in general, and for channel function and formation of dendritic spines in hippocampal pyramidal cells in particular (Ito et al., 1997; Cull-Candy et al., 2001; Kim et al., 2005; Akashi et al., 2009). Transgenic overexpression of Grin2b in the forebrain of mice, and in the cortex and hippocampus of rats results in an increased activation of the NMDARs, with mice and rats showing a superior performance in various tests of learning and memory (Tang et al., 1999; Wang et al., 2009). These findings point toward GRIN2B as a susceptibility gene for complex traits/disorders in humans. Nevertheless, although genetic contributions are clearly relevant to the etiology of these phenotypic expressions, it is widely recognized that additional factors – whose nature cannot be immediately identified as genetic – also act as powerful predictors for human cognition and neurodevelopmental disorders, such as DD. A large range of factors, that is pre/ perinatal (Gilger et al., 1992; Fried et al., 1997; Bowen et al., 2002; Batstra et al., 2003; Samuelsson et al., 2006; Michaelsen et al., 2009; van Baar et al., 2009; Hoque et al., 2012), sociodemographic, and familial (Fergusson and Lynskey, 1993; Fergusson and Woodward, 1999; Melekian, 2001; Hoff and Tian, 2005), seems to impinge on human cognition and DD (Davis et al., 2001; Gayán and Olson, 2001, 2003; Byrne et al., 2002; Petrill et al., 2006; Harlaar et al., 2007), and may contribute toward increasing the likelihood of developing this disorder. Given the above evidence, it thus seems appropriate to complement genomic investigation of such complex traits by taking into account the gene-by-environment interaction (G × E), which is a specific form of interplay, whereby genetic susceptibility conferred by a specified allele is modulated by a measured environmental factor (Rutter et al., 2006).
Here, we sought to resolve which aspects of the phenotypic profile of DD were unambiguously attributable to GRIN2B effects by exploring the contribution of four SNPs that have been shown previously to contribute toward the genetic background of DD (i.e. rs1012586G/C; rs2268119A/T; rs2216128T/C; rs2192973C/T; Ludwig et al., 2009) and two markers (i.e. rs5796555 − /A and rs11609779C/T) close enough to markers rs1012586G/C and rs2216128T/C, respectively, to be amplified and sequenced with them, to DD, defined as a categorical trait, and key DD-related phenotypes by association analyses in a sizeable sample of Italian nuclear families. Finally, we tested the role of the selected GRIN2B markers and a set of commonly described environmental moderators using a test for G × E interaction in a sib pairbased association analysis of quantitative traits (van der Sluis et al., 2008; Mascheretti et al., 2013a).
Materials and methods The protocol was approved by the Scientific Institute ‘Eugenio Medea’ Ethics and Scientific Board. Sample
This study is part of an ongoing project on the genetics of reading disabilities at the Department of Child Psychiatry and Rehabilitation Centre at the Scientific Institute ‘Eugenio Medea’, Bosisio Parini, Italy, and at the Centro Regionale di Riferimento per i Disturbi dell’Apprendimento – CRRDA (‘Regional Reference Center for the Specific Learning Disability’, ULSS 20, Verona, Italy) (Mascheretti et al., 2013a, 2013b, 2014). To date, 466 unrelated nuclear families of probands with DD (738 offspring) have been recruited; except for 47 families that had only one parent available, all parents were represented, yielding a total sample of 1623 individuals, all of Italian ancestry. The ascertainment scheme has been reported in detail elsewhere (Marino et al., 2003). Briefly, nuclear families were recruited if probands met the criteria for DD according to the DSM-IV (American Psychiatric Association, 1994). After parental informed consent, offspring underwent an extensive medical assessment and a battery of tests, which evaluate text, word, nonword reading (Cornoldi, 1995; Sartori et al., 1995; Cornoldi et al., 1998), writing-under-dictation of word, nonword and sentences-containing-homophones (Sartori et al., 1995), forward/backward digit spans (Reynolds and Bigler, 1994), phonemic elision and blending (Cossu et al., 1988), mathematics abilities (Cornoldi et al., 2003; Cornoldi and Lucangeli, 2004), and the Wechsler Intelligence Scale for Children, Revised (WISC-R) (Wechsler, 1981) or the Wechsler Intelligence Scale for Children, 3rd ed. (WISC-III) (Wechsler, 2006). For all tests, standardized scores on the Italian population are provided (for an extensive description of tests, see Supplemental digital content 1, http://links.lww.com/PG/
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
GRIN2B and susceptibility to dyslexia Mascheretti et al. 11
A119). Sibs were administered only two subtests of the intelligence scale, that is vocabulary and block design, which shows a high correlation (r) with, respectively, verbal IQ (r = 0.82; Wechsler, 1981, 2006) and performance IQ (r = 0.73; Wechsler, 1981, 2006). The criteria used to define the probands’ affection status were as follows: (a) a performance on a timed text-reading test at least 2 SD below the expected grade level mean on either accuracy or speed; or (b) an absolute score at least 2 SD below the expected grade level mean on accuracy or speed in a reading list of unrelated words or nonwords; and (c) IQ more than 84. Siblings were included if they were fully biological, older than 6 years and younger than 18 years, if they had no history of neurological and sensorial disorders, and if the mean score of vocabulary and block design subtests was above 7, irrespective of their reading performance. Blood or mouthwash samples were obtained from all offspring and their biological parents.
Phenotypes
We considered eight phenotypes in the single-marker transmission disequilibrium test (TDT) analyses: (1) DD as a discrete trait. (2) READING, as measured by averaging speed (s) and accuracy (expressed in number of errors) gradestandardized scores in text, word, and nonword reading tasks (Cornoldi, 1995; Sartori et al., 1995; Cornoldi et al., 1998) as the mean bivariate correlation among the above-mentioned tasks was considerable (r = 0.450). (3) SPELLING, as measured by averaging the accuracy (expressed in number of errors), grade-standardized scores in writing-under-dictation word, nonword, and sentences-containing-homophones (bivariate correlation among the above-mentioned tests, r = 0.563; Sartori et al., 1995). (4) Performance IQ, as measured by the WISC-R or the WISC-III. For siblings, the ‘block design’ subtest was used to provide a prorated, performance IQ score (Wechsler, 1981, 2006). (5) Auditory STM, as measured by averaging agenormed scores in the forward and backward digit spans (bivariate correlation among the abovementioned spans, r = 0.701; Reynolds and Bigler, 1994). (6) Phonemic elision (ELISION; Cossu et al., 1988). (7) Phonemic blending (BLENDING; Cossu et al., 1988). (8) Mathematical abilities (Cornoldi et al., 2003; Cornoldi and Lucangeli, 2004), as measured by a principalcomponent-analysis factor that accounted for 53% of the total test variance among mental and written calculation, number dictation, and numerical facts (mean factor loadings = 0.64 ± 0.08; MATHEMATICS).
Of the total sample (n = 466; see the Sample section), (i) 324 families (538 offspring) had complete measures of READING, SPELLING, performance IQ, auditory STM (group 1), and (ii) 174 families (223 offspring) had complete measures of ELISION, BLENDING and MATHEMATICS (group 2). Group 1 and group 2 overlapped with respect to the phenotypic data available.
Environmental data collection
Selection of the environmental factors has been described extensively elsewhere (Mascheretti et al., 2013a, 2013b). Briefly, using a large case–control sample, we removed all variables with an at-risk category frequency under 5% and collapsed those variables that were highly correlated to obtain more concise and comprehensive variables (Mascheretti et al., 2013a, 2013b). Environmental factors included in further analyses were as follows: (1) Maternal smoking – hereafter smoke. (2) Risk of miscarriage during pregnancy – hereafter miscarriage. (3) Birth weight. (4) Breast feeding. (5) Parental age, as an average of the father and the mother’s age at child’s birth. (6) Socioeconomic status, defined by parental employment. (7) Parental education, as an average of the father and the mother’s educational qualifications. Dichotomous responses (yes/no) were coded with ‘0’ when answers were ‘no’ and ‘1’ when answers were ‘yes’, except for breast feeding, which was coded as ‘0’ if child had been breastfed. The employment response was coded according to the Hollingshead nine-point scale (Hollingshead, 1975). A score (from 10 to 90) was assigned to each job; the higher of two scores was used when both parents were employed. The educational qualification response was scored according to a ninepoint ordinal scale on the basis of the Italian school system (range between 10, corresponding to fifth-grade elementary school, and 90, equivalent to a postdoctoral degree). One-hundred and seventy-eight families (144 pairs, 31 triplets, two with four siblings, and one with five siblings, 394 offspring) had at least one sibling and complete environmental data (group 3); in this latter group, complete phenotypes were available for READING, SPELLING, performance IQ, and auditory STM, whereas only 60 families had complete data for ELISION, BLENDING, and MATHEMATICS, which were therefore excluded from G × E analyses because of limited power. As for group 2, group 1 and group 3 overlapped with respect to the phenotypic data available. Descriptive statistics of the environmental variables are outlined in Supplemental digital content 2 (http://links. lww.com/PG/A120). All environmental factors, except
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
12 Psychiatric Genetics 2015, Vol 25 No 1
SMOKE, MISCARRIAGE, breast feeding, and parental marital status, were z-transformed on the basis of the sample norms to avoid a scaling effect in further analyses.
Fig. 1
Genotype error checking was completed in PEDSTATS (Wigginton and Abecasis, 2005) and inconsistent genotypes were zeroed out and were not considered for further analysis. Families were excluded in the case of multiple inconsistent genotypes. Allelic frequencies and Hardy– Weinberg equilibrium for the markers under consideration were calculated in parents (Supplemental digital content 3, http://links.lww.com/PG/A121). No significant deviations from allele frequencies reported in Ludwig et al. (2009) were found (Supplemental digital content 3, http://links. lww.com/PG/A121). For all the six genotyped SNPs, P-values for deviation from the Hardy–Weinberg equilibrium were not lower than 0.008 (P = 0.05/6; Supplemental digital content 3, http://links.lww.com/PG/A121), and thus no SNPs were excluded. The linkage disequilibrium structure of GRIN2B was analyzed using only the parental genotypes; linkage disequilibrium was extracted and plotted in Haploview 4.0 (Fig. 1). As shown in Fig. 1, even if markers rs5796555 − A and rs11609779C/T were typed as a byproduct of rs1012586G/C and rs2216128T/C, respectively, they were not excluded from further analysis as they did not show correlation with any of other genotyped SNPs. Statistical analysis Genetic association analysis
Genetic association was investigated using a family-based association test, that is, the quantitative transmission disequilibrium test (QTDT, version 2.5.1) as modeled by Abecasis et al. (2000), which allows to control for
3
39
20 8
Block 1 (13 kb) 4 5
3
6 5
6
25
3
rs2192973
2
rs11609779
rs2268119
1
rs2216128
rs1012586
Amplification and sequencing of four portions of the human GRIN2B gene allowed typing of the following SNPs: rs5796555 − /A; rs1012586G/C; rs2268119A/T; rs2216128T/C; rs11609779C/T; rs2192973C/T. In particular, we chose to genotype the four SNPs that in a previous study were found to be associated significantly with short-term memory in a German DD sample (i.e. rs1012586, rs2268119, rs2216128, and rs2192973; Ludwig et al., 2009), suggesting that variations in GRIN2B may contribute toward the genetic background of specific cognitive processes that are correlated to DD. Markers rs5796555 − /A and rs11609779C/T are close enough to markers rs1012586G/C and rs2216128T/C, respectively, to be amplified and sequenced with them. Amplifications were performed in 10-μl reactions using JumpStart Red ACCUTaq LA DNA polymerase (Sigma, St. Louis, Missouri 63103, USA) and the following protocol: 30 s at 96°C, 35 cycles of 15 s at 94°C/20 s at 58°C/30 s at 68°C, and a 5 min final elongation time. Sequencing reactions were performed using a Big Dye Terminator Cycle Sequencing kit (Applied Biosystems, St. Louis, Missouri 63103, USA) and run on an ABI Prism 3130xl Genetic Analyzer. (Primers are available on request from the authors.)
rs5796555
Genotyping
91
3 26
1 4
1 4
Haploview plot showing pairwise linkage disequilibrium (r2 values) for six SNPs of GRIN2B on the basis of parents’ genotypes. Numbers in squares are r2 values.
population stratification bias and for the analyses of quantitative traits. Single-marker TDT analyses were carried out for DD as a discrete trait using the ‘-ad’ option, which allows to test using only affected individuals. Quantitative traits were analyzed using the ‘-wega’ option, which allows to adopt commonly used variance components, that is, the environmental variance (e), polygenic variance (g), and additive major locus (a). Genetic findings are not presented for the entire set of families because of missing data and because TDTs require individuals to have heterozygous ancestors in the pedigree; thus, offspring of homozygous parents are uninformative (Abecasis et al., 2000). We calculated the genetic association for the total sample. Moreover, as follow-up association signals over different severity groups may confirm a possible true contribution of the selected SNPs to DD and DD-related performance (Ludwig et al., 2009), we tested the genetic association for a subsample selected by severity, that is, by selecting only the nuclear families in which at least one offspring scored less than or equal to 2.50 SD below the general population mean on either accuracy or speed in either text-reading, or word-reading, or nonword reading tasks (n = 292). Tables 1 and 2 show the descriptive statistics of all traits in groups 1 and 2 of the total sample and in the selected-by-severity subsample, respectively. Only empirical P-values are reported, which are computed from 10 000 Monte-Carlo permutations by entering all
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Measures are in SD units, relative to the age-appropriate Italian population norm. Auditory STM, the average between forward/backward digit spans; MATHEMATICS, the regressed factor scores obtained by the PCA; READING, the average between text-reading, word-reading, and nonword reading tasks (both accuracy and speed); SPELLING, the average between word, nonword spelling and orthographic choice tasks. a Age was expressed in years. b Percentage of males was reported. c For siblings, performance intelligence quotient (IQ) represents an estimated score obtained from the ‘Block Design’ subtest as described in the text. d These data were collected on 174 nuclear families.
− 0.34 17.29 5.63 1.28 11.92 1.29 2.61 0.09 − 3.20 − 2.21 − 0.04 − 2.98 − 1.14 − 1.42 (17.37) (1.89) (2.10) (1.04) (1.34) (2.57) (0.82) 109.58 − 0.95 − 0.99 − 0.36 − 0.71 − 2.80 0.51 157 1.03 0.88 2.50 0.59 − 0.17 1.55 65 − 15.43 − 11.30 − 3.33 − 7.74 − 12.20 − 2.49 1.22 6.01 4.42 0.39 1.29 − 0.10 0.57 0.59 − 1.93 − 2.00 0.87 − 1.13 − 0.75 − 0.90 (10.86) (1.72) (3.20) (1.15) (1.40) (2.80) (1.00) 105.58 − 2.53 − 2.63 − 0.28 − 1.22 − 3.66 − 0.14 160 0.09 1.29 3.17 0.88 − 0.17 1.51 80 − 11.52 − 16.30 − 3.33 − 6.25 − 11.44 − 3.40 0.28 6.93 5.63 0.79 3.26 0.16 0.71 0.37 − 1.83 − 2.16 0.50 − 1.57 − 0.85 − 0.96 (15.11) (1.95) (2.92) (1.11) (1.39) (2.75) (1.00) 107.38 − 1.90 − 1.96 − 0.32 − 1.06 − 3.40 0.00 160 1.03 1.29 3.17 0.88 − 0.17 1.55 65 − 15.43 − 16.30 − 3.33 − 7.74 − 12.20 − 3.40 Performance IQ READING SPELLING Auditory STM ELISIONd BLENDINGd MATHEMATICSd
Kurtosis Skew. Mean (SD) Maximum Minimum
Maximum
Mean (SD)
Skew.
Kurtosis
Minimum
Siblings Age = 12.55 (± 3.92)a; sex = 51.5%b
Kurtosis Skew. Mean (SD) Maximum
where τi is the family-specific intercept, ab and aw are the estimated between-family and within-family additive genetic effects of the marker, according to the orthogonal decomposition, e represents the effects of the environmental risk factor, ibg and iwg represent the between-family and within-family effects of the interaction of genotype g and the environmental risk factor, and εij is the residual term (van der Sluis et al., 2008). As such, iwg becomes an estimate of the G × E effect inasmuch as it indicates the change in the allelic variant marker association with the phenotype across the different ecological niches, and it represents the variation of SD units of the z-score on the neuropsychological composite. For instance, a positive value represents an increase in the effect of each additional
Minimum
yijg¼ ti þab Abi þaw Awij þeEi þibg Eij Abi þiwg Eij Awij þeij ;
c
The complete description of this statistical model has been reported in detail elsewhere (see Supplemental digital content 4, http://links.lww.com/PG/A122; Mascheretti et al., 2013a). Briefly, this is an extension of the Fulker et al. (1999) maximum likelihood variance components analysis of quantitative traits in sib-pairs data that incorporates environmental main effects plus G × E effects (van der Sluis et al., 2008). The association effect is orthogonally decomposed here into between-family (b) and withinfamily (w) effects, that is, the expectation of each sib genotype conditional on family genotype data and the deviation from this expectation for each offspring, respectively. To model the interaction effect, we used the sibling-specific value for each environmental risk variable; the phenotypic score (i.e. the observed phenotypic score y for participant j from family i with genotype g) is then modeled as follows:
Probands Age = 10.45 (2.77)a; sex = 72.7%b
To explore the combined role of candidate genes markers and measured environmental factors on performance IQ and DD-related neuropsychological skills (i.e. READING, SPELLING, and Auditory STM), we analyzed the effects of G × E through a general test for G × E interaction in sib pair-based association analysis of quantitative traits (van der Sluis et al., 2008). Table 3 shows the descriptive statistics of the selected neuropsychological traits in group 3.
Total sample Age = 11.36 (3.48)a; sex = 64.9%b
Gene-by-environment interaction analysis
Table 1
the selected markers (see the Materials and methods and Genotyping sections) and the above-described phenotypes (see the Materials and methods and Phenotypes sections) at the same time. Bonferroni correction for multiple testing was not applied because it would have been too conservative (Deffenbacher et al., 2004; Francks et al., 2004; Cope et al., 2005; Meng et al., 2005; Schumacher et al., 2006; Brkanac et al., 2007; Marino et al., 2012). Thus far, we decided to adjust the significance levels by the false discovery rate (FDR) method (Storey, 2002) applied to the seven neuropsychological traits analyzed for each marker, separately for each SNP.
Descriptive statistics of the neuropsychological measures in groups 1 and 2 of the total sample (n = 324 and 174, respectively) with complete phenotypic and genetic information
GRIN2B and susceptibility to dyslexia Mascheretti et al. 13
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Measures are in SD units, relative to the age-appropriate Italian population norm. Auditory STM represents the average between single number forward/backward spans; MATHEMATICS, the regressed factor scores obtained by the PCA; READING, the average between text-reading, word-reading and nonword reading tasks (both accuracy and speed); SPELLING, the average between word, nonword spelling and orthographic choice tasks. a Age was expressed in years. b Percentage of males was reported. c For siblings, performance intelligence quotient (IQ) represents an estimated score obtained from the ‘Block Design’ subtest, as described in the text. d These data were collected on 150 nuclear families.
− 0.38 15.59 5.04 1.30 11.11 1.99 2.46 − 0.11 − 3.03 − 2.10 0.03 − 2.91 − 1.35 − 1.41 150 1.03 0.88 2.50 0.59 − 0.17 1.55 65 − 15.43 − 16.30 − 3.33 − 7.74 − 12.20 − 3.40
160 1.03 1.29 3.17 0.59 − 0.17 1.55
107.58 − 2.12 − 2.07 − 0.31 − 1.13 − 3.36 − 0.03
(14.73) (1.99) (3.04) (1.11) (1.44) (2.78) (1.01)
0.25 − 1.76 − 2.09 0.54 − 1.52 − 0.85 − 0.95
0.13 6.65 5.08 0.70 2.99 0.06 0.63
80 − 11.52 − 16.30 − 3.33 − 6.25 − 11.44 − 3.40
160 − 0.11 1.29 3.17 0.59 − 0.17 1.51
105.66 − 2.76 − 2.74 − 0.28 − 1.31 − 3.68 − 0.17
(12.61) (1.71) (3.35) (1.16) (1.42) (2.83) (1.01)
0.58 − 2.00 − 1.91 0.82 − 1.07 − 0.69 − 0.88
1.13 6.20 3.86 0.28 1.14 − 0.33 0.47
65 − 15.43 − 11.30 − 3.33 − 7.74 − 12.20 − 2.49
109.98 − 1.12 − 1.08 − 0.36 − 0.73 − 2.67 0.44
(16.74) (1.98) (2.18) (1.05) (1.40) (2.57) (0.84)
Kurtosis Skew. Mean (SD) Maximum Skew. Mean (SD) Maximum Skew. Mean (SD) Maximum Minimum
Total sample Age = 11.40 (3.49)a; sex = 62.5%b
Kurtosis
Minimum
Probands Age = 10.58 (2.83)a; sex = 70.5%b
Kurtosis
Siblings Age = 12.50 (± 3.96)a; sex = 52.2%b
Minimum
c
Performance IQ READING SPELLING Auditory STM ELISIONd BLENDINGd MATHEMATICSd
Table 2
Descriptive statistics of the neuropsychological measures in the selected-by-severity subsample (n = 292) with complete phenotypic and genetic information
14 Psychiatric Genetics 2015, Vol 25 No 1
transmission of the minor allele on the neuropsychological composite z-score when environmental variable increases. A negative value of iwg represents a decrease of the same effect. The permutation procedure was repeated 1000 times for each analysis. We decided to adjust the significance levels using the FDR method (Storey, 2002) applied to the 28 tests performed for each marker (7 environmental variable × 4 phenotypes), separately for each marker (Mascheretti et al., 2013a). Sex was taken into account in the extended equation because probands’ sex ratio (males : females) in our sample was almost 3 : 1, and it may imply differences in the mean scores between males and females. Moreover, as simulation studies showed specific situations in which dichotomized variables performed as well as or better than the original quantitative factors, we subsequently decided to dichotomize raw scores of quantitative environmental variables (i.e. birth weight, parental age, socioeconomic status, and parental education; see the Materials and methods and the Environmental data collection sections) according to well-defined cut-off points (for more information, see Supplemental digital content 4, http://links.lww.com/PG/A122).
Results Genetic association analyses with developmental dyslexia and developmental dyslexia-related quantitative traits
Significant associations were found between DD as a discrete trait and the minor alleles ‘A’ and ‘G’ of the markers rs5796555 − /A and rs1012586G/C, respectively, in the total sample (informative families = 331, T = − 2.04, SD = 8.63, nominal P-value = 0.043, and informative families = 347, T = − 1.99, SD = 8.65, nominal P-value = 0.047, respectively). In the selected-by-severity subsample DD as a discrete trait was associated significantly with the minor allele ‘A’ of the marker rs5796555 − /A (informative families = 227, T = − 1.99, SD = 6.65, nominal P-value = 0.048), whereas only a trend toward significance was observed for the rs1012586G/C marker (informative families = 234, T = − 1.79, SD = 6.84, nominal P-value = 0.076). Tables 4 and 5 show both the empirical P-values obtained after implementing 10 000 Monte-Carlo permutations and the q-values after FDR correction. Markettrait associations for READING and SPELLING were nonsignificant in both group 1 and in the selected-by-severity subsample (Table 4). Performance IQ was associated significantly with the minor alleles ‘T’, ‘C’, and ‘T’ of markers rs2268119A/T (χ2 = 6.44; nominal P-value = 0.011; empirical P-value = 0.009; q-value = 0.004; 224 informative families; genetic effect = − 4.714; Table 4), rs2216128T/C (χ2 = 12.38; nominal P-value < 0.001; empirical P-value = 0.001; q-value = 0.007; 227 informative families; genetic effect = − 6.178; Table 4), and rs2192973C/T (χ2 = 10.04; nominal P-value = 0.002; empirical P-value = 0.001; q-value = 0.007; 211 informative families; genetic effect = − 3.460: Table 4),
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
GRIN2B and susceptibility to dyslexia Mascheretti et al. 15
Descriptive statistics of the neuropsychological measures in group 3 of the total sample (n = 178) with complete phenotypic and genetic information
Table 3
Probands (n = 178) Siblings (n = 216) Affected (n = 265) Age = 10.52 (2.81)a; Age = 12.43 (3.81)a; Age = 13.17 (3.12)a; sex = 69.1%b sex = 52.8%b sex = 64.2%b c
Total sample (n = 394) Age = 11.56 (3.51)a; sex = 60.5%b Minimum Maximum Performance IQd READING SPELLING MEMORY NUMBERS
44 − 15.43 − 16.30 − 3.33
160 1.03 1.16 3.00
Mean (SD) 108.41 − 1.66 − 1.75 − 0.42
(16.36) (2.00) (2.92) (1.04)
Not-affected (n = 120) Age = 12.84 (3.94)a; sex = 52.5%b
Skew.
Kurtosis
Mean
SD
Mean
SD
Mean
SD
Mean
SD
0.12 − 2.21 − 2.27 0.43
0.58 8.89 6.11 1.39
107.26 − 2.42 − 2.59 − 0.46
14.80 1.77 3.38 1.06
109.21 − 1.01 − 1.04 − 0.39
17.36 1.96 2.25 1.03
106.53 − 2.39 − 2.40 − 0.53
15.09 1.98 3.22 1.00
111.96 − 0.06 − 0.37 − 0.23
18.00 0.71 1.35 1.08
Measures are in SD units, relative to the age-appropriate Italian population norm. MEMORY NUMBERS, the average between single number forward/backward spans; READING, the average between text-reading, word-reading, and nonword reading tasks (both accuracy and speed); SPELLING, the average between word, nonword spelling, and orthographic choice tasks. a Age was expressed in years. b Percentage of males was reported. c The affection status was assigned according to the criteria outlined in the text. d For siblings, performance intelligence quotient (IQ) represents an estimated score obtained from the ‘Block Design’ subtest, as described in the text.
respectively, in group 1. In the selected-by-severity subsample, performance IQ was associated significantly with the minor allele ‘C’ of marker rs2216128T/C (χ2 = 8.31; nominal P-value = 0.004; empirical P-value = 0.004; q-value = 0.024; 140 informative families; genetic effect = − 6.509; Table 4), whereas only a trend toward significance was found for the rs2268119A/T and rs2192973C/T markers (Table 4). Auditory STM yielded a significant association with the minor allele ‘A’ of the rs5796555 − /A marker in group 1 (χ2 = 5.15; nominal P-value = 0.023; empirical P-value = 0.040; q-value = 0.037; 244 informative families; genetic effect = − 0.240; Table 4). ELISION was associated significantly with the minor allele ‘G’ of the rs1012586G/C marker in the selected-by-severity sample (χ2 = 3.37; nominal P-value = 0.068; empirical P-value = 0.037; q-value = 0.041; 118 informative families; genetic effect = − 0.427; Table 5). BLENDING and MATHEMATICS did not yield any significant association with any markers (Table 5) in both group 2 and the selected-by-severity subsample.
Gene-by-environment interaction analysis
We analyzed G × E interaction effects between six markers and seven environmental factors using the general test for G × E interaction upon READING, SPELLING, performance IQ, and auditory STM composites in group 3. Although we found G-E combinations whose empirical P-values of the iwg term was significant, none survived after FDR correction (Supplemental digital content 5, http://links.lww.com/PG/A123). With respect to the investigation of the sex effect, no empirical P-values were significant (Supplemental digital content 5, http://links. lww.com/PG/A123). Finally, we analyzed G × E interaction effects between six markers and the four dichotomized quantitative environmental factors (see the Materials and methods, the ‘Statistical analysis’, the Gene-by-environment interaction analysis sections, and Supplemental digital
content 4, http://links.lww.com/PG/A122) using the general test for G × E interaction upon the same neuropsychological phenotypes (i.e. READING, SPELLING, performance IQ, and auditory STM) in group 3. We found G × E combinations whose empirical P-values of the iwg term were significant and that showed 33% overlap with the results obtained by implementing quantitative variables, but none survived after FDR correction (Supplemental digital content 6, http://links.lww.com/PG/ A124). No significant sex effect was found (Supplemental digital content 6, http://links.lww.com/PG/A124).
Discussion The ability to read relies on a complex, highly integrated, large-scale network of different cognitive processes. Recently, research applied quantitative cognitive measures to dissect the heterogeneous DD phenotype with respect to the genetic contribution (Marlow et al., 2003; Paracchini et al., 2007; Schulte-Körne et al., 2007). This study was primarily designed to replicate previous findings of the involvement of the GRIN2B gene in DD and its related neuropsychological phenotypes in a sizable sample of Italian nuclear families ascertained for DD. We therefore investigated both linear and G × E interplay effects on DD and DD-related quantitative phenotypes. QTDT analyses yielded significant associations between markers spanning within GRIN2B and DD as a categorical trait and its related quantitative phenotypes, that is, performance IQ, auditory STM, and phonemic elision. These findings suggest that GRIN2B can account for not only part of the disabilities proper of DD, defined as a categorical trait, but also for part of the reduced skills in key DD-related cognitive domains that are observable in these children. Indeed, DD is a heterogeneous disorder in which various phenotypic dimensions are involved, including phonological processing and memory processes (Gabrieli, 2009). As such, our
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
262
213
244
206
311
217
324
159
3.41 (341)
5.15 (400)
0.05 (326)
0.95 (499)
0.68 (350)
0.55 (529)
0.02 (247)
0.35 (415)
χ2 (d.f.)
0.040 0.037b 0.078 0.546b
0.395 0.070b 0.790 0.900b
0.461 0.070b 0.434 0.900b
0.497 0.070b 0.900 0.900b
Empirical P-value
229
260
216
326
227
342
158
265
Informative families
2.31 (342)
2.48 (399)
0.50 (327)
0.00 (499)
3.93 (351)
2.41 (529)
0.18 (247)
1.07 (413)
χ2 (d.f.)
rs1012586G/C
0.114 0.132b 0.134 0.050b
0.981 0.386b 0.378 0.070b
0.209 0.132b 0.135 0.050b
0.239 0.132b 0.653 0.104b
Empirical P-value
184
212
167
261
184
282
134
224
Informative families
0.30 (334)
0.21 (388)
0.76 (315)
0.31 (484)
0.96 (339)
0.80 (514)
4.23 (238)
6.44 (401)
χ2 (d.f.)
rs2268119A/T
0.574 0.038b 0.498 0.185b
0.541 0.038b 0.339 0.163b
0.418 0.038b 0.358 0.163b
0.009 0.004b 0.055 0.122b
Empirical P-value
191
221
174
263
187
282
140
227
Informative families
0.65 (333)
0.14 (388)
0.06 (314)
0.09 (484)
0.00 (338)
0.63 (514)
8.31 (236)
12.38 (400)
χ2 (d.f.)
rs2216128T/C
0.776 0.776b 0.567 0.689b
0.763 0.776b 0.810 0.820b
0.424 0.742b 0.964 0.837b
0.001 0.007b 0.004 0.024b
Empirical P-value
164
195
162
246
187
263
128
208
Informative families
0.05 (333)
0.02 (389)
3.22 (315)
2.02 (486)
0.00 (338)
0.02 (516)
002 (237)
0.48 (402)
χ2 (d.f.)
rs11609779C/T
0.862 0.887b 0.790 0.964b
0.121 0.847b 0.024 0.168b
0.887 0.887b 0.964 0.964b
0.487 0.852b 0.881 0.964b
Empirical P-value
173
203
159
244
171
262
128
211
Informative families
0.59 (333)
0.11 (387)
0.01 (314)
0.01 (483)
0.06 (338)
0.11 (513)
6.73 (238)
10.04 (401)
χ2 (d.f.)
rs2192973C/T
0.810 0.941b 0.615 0.796b
0.941 0.941b 0.932 0.862b
0.731 0.941b 0.813 0.862b
0.001 0.007b 0.009 0.058b
Empirical P-value
Auditory STM, the average between forward/backward digit spans; READING, the average between text-reading, word-reading, and nonword reading tasks (both accuracy and speed); SPELLING, the average between word, nonword spelling, and orthographic choice tasks. a For siblings, performance intelligence quotient (IQ) represents an estimated score obtained from the ‘Block Design’ subtest, as described in the text. b FDR correction’s q-values are bold if ≤ 0.05.
Severity sample
Auditory STM Total sample
Severity sample
SPELLING Total sample
Severity sample
READING Total sample
Severity sample
Informative families
rs5796555 − /A
GRIN2B marker
Market-trait association empirical P-values in both group 1 of the total sample (n = 324) and the selected-by-severity subsample (n = 292) for GRIN2B markers
Performance IQa Total sample
Table 4
16 Psychiatric Genetics 2015, Vol 25 No 1
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
0.041 0.144a 0.101 0.327a 3.98 (225)
2.61 (195) 103
124
0.479 0.852a 0.798 0.964a 0.07 (195)
0.53 (227) 3.38 (226)
2.85 (195)
139
118
0.65 (226) 117 0.37 (238)
MATHEMATICS, the regressed factor scores obtained by the PCA. a FDR correction’s q-values are bold if
