TL;DR
This study investigated whether socioeconomic status (SES) influences the genetic predisposition to dyslexia and reading skills. Using polygenic indices derived from a large dyslexia GWAS, researchers found that while both genetics and SES independently predict reading ability, they do not interact to affect outcomes in Australian and UK cohorts. This suggests that SES does not moderate the expression of genetic risk for dyslexia in these populations, contrary to some US-based findings.
What Problem Does This Paper Address?
Reading and spelling difficulties, like dyslexia, significantly impact life outcomes. While both genetic and environmental factors contribute to these abilities, there's a long-standing question about whether genetic predispositions are amplified or buffered by environmental factors, particularly socioeconomic status (SES). Previous research, primarily from the US, suggested a gene x SES interaction, where heritability of reading skills might be higher in higher SES environments. However, the generalizability of these findings to other countries with different educational and social systems was unclear. This paper addresses this gap by examining the gene x SES interaction for dyslexia and reading skills in Australian and UK cohorts, using a more direct measure of genetic predisposition (polygenic indices) than previous twin studies.
Methods
The study employed a polygenic risk score (PRS) approach. First, they utilized summary statistics from a large genome-wide association study (GWAS) on dyslexia to create polygenic indices (PGI) for each participant. These PGIs represent an individual's estimated genetic predisposition to dyslexia based on thousands of genetic variants. The researchers then used linear regression models to assess the independent effects of the PGI and family SES on reading and spelling performance. Crucially, they tested for an interaction effect between the PGI and SES, hypothesizing that SES would moderate the genetic influence on reading skills. In the Australian cohort, linear mixed-effects models were used to account for family relationships.
Datasets
The study utilized data from two large, established cohorts:
- Australian Cohort (Brisbane Adolescent Twin Study - BATS): This cohort included 1315 adolescents.
- UK Cohort (1958 National Child Development Study - NCDS): This cohort provided data from two time points: age 7 (N=5461) and age 16 (N=4306).
Genetic data for these cohorts were derived from DNA samples, and phenotypic data for reading and spelling were collected using standardized tests and teacher ratings. SES was measured using parental occupation and education information.
Key Insights and Ideas
- No Gene x SES Interaction Found: The most significant finding is the absence of a gene x SES interaction effect on reading skills in both Australian and UK cohorts. This challenges previous US-based findings and suggests that the way genes influence reading ability is not significantly modified by socioeconomic background in these populations.
- Independent Effects of Genes and SES: Despite the lack of interaction, both the dyslexia PGI and family SES independently predicted reading and spelling performance. This reinforces the understanding that both genetic predisposition and environmental factors play crucial roles.
- Potential for Gene-Environment Correlation: Interestingly, in both cohorts, individuals in lower SES groups tended to have higher average dyslexia PGIs. This suggests a potential gene-environment correlation, where family SES might reflect parental genetic factors that are then passed on, rather than SES directly moderating the genetic effect on the child's reading.
- Country-Specific Differences: The results support the idea that gene-environment interactions can be country-specific, likely due to differences in educational systems, social mobility, and the strength of SES's relationship with educational quality.
Key Results
- Australia: In the Australian cohort, SES showed a significant positive effect on reading and spelling performance (a one standard deviation increase in SES was associated with a 0.22 standard deviation increase in reading performance, p<0.001). The dyslexia PGI showed a significant negative effect (a one standard deviation increase in PGI was associated with a 0.20 standard deviation decrease in reading performance, p<0.01). However, the interaction term between PGI and SES was not significant (p=0.29), with low statistical power (17%) to detect it.
- UK (Age 7): In the UK cohort at age 7, SES had a significant positive effect (a one standard deviation increase in SES was associated with a 0.26 standard deviation increase in reading performance, p<0.001). The dyslexia PGI had a significant negative effect (a one standard deviation increase in PGI was associated with a 0.11 standard deviation decrease in reading performance, p<0.001). The interaction term was not significant (p=0.16), with a post-hoc power of 30% to detect an effect size of 0.02.
- UK (Age 16): At age 16 in the UK cohort, SES again showed a significant positive effect (a one standard deviation increase in SES was associated with a 0.33 standard deviation increase in reading performance, p<0.001). The dyslexia PGI had a significant negative effect (a one standard deviation increase in PGI was associated with a 0.09 standard deviation decrease in reading performance, p<0.001). The interaction term was not significant (p=0.19), with a post-hoc power of 28% to detect an effect size of 0.02.
- Overall, polygenic indices and SES together explained between 8% and 11.6% of the variance in reading and spelling performance across the different samples and ages.
Limitations and Future Work
The authors acknowledge several limitations:
- Limited Predictive Power of PGI: The dyslexia PGI, while derived from a large GWAS, still explains a relatively small proportion of variance in reading skills. This limits the power to detect subtle gene x SES interactions. Future GWAS with larger sample sizes and more comprehensive measures of reading ability (beyond dyslexia diagnosis) could improve predictive power.
- Self-Selected Sample: The 23andMe data used for the GWAS is from a self-selected consumer genetics sample, which may not be fully representative of the general population.
- Proxy Measure: The study used a dyslexia PGI as a proxy for general reading ability, which might not capture all nuances of quantitative reading skills.
- UK Cohort Specifics: The UK NCDS cohort, particularly the genotyped subsample, might be less representative, and the timing of testing at age 7 could be influenced by recent schooling.
- Broad SES Measure: Family SES is a broad measure. Future research could benefit from more granular environmental measures, such as specific literacy environments (e.g., school quality, home literacy practices), to better understand potential gene-environment interactions and inform educational policies.
Future work should focus on using more contemporary data to account for societal changes, employing more precise genetic and environmental measures, and exploring gene x environment interactions in more heterogeneous populations.
Summary updated