This is an Open Access document downloaded from ORCA, Cardiff University's institutional repository: http://orca.cf.ac.uk/113166/ This is the author s version of a work that was submitted to / accepted for publication. Citation for final published version: Escott-Price, Valentina, Bracher-Smith, Matthew, Menzies, Georgina, Walters, James, Kirov, George, Owen, Michael J. and O'Donovan, Michael C. 2018. Genetic liability to schizophrenia is negatively associated with educational attainment in UK Biobank. Journal of Molecular Psychiatry file Publishers page: Please note: Changes made as a result of publishing processes such as copy-editing, formatting and page numbers may not be reflected in this version. For the definitive version of this publication, please refer to the published source. You are advised to consult the publisher s version if you wish to cite this paper. This version is being made available in accordance with publisher policies. See http://orca.cf.ac.uk/policies.html for usage policies. Copyright and moral rights for publications made available in ORCA are retained by the copyright holders.
Genetic liability to schizophrenia is negatively associated with educational attainment in UK Biobank Valentina Escott-Price 1*, PhD, Matthew Bracher-Smith 1, BSc, Georgina Menzies 1, PhD, James Walters 1, PhD, George Kirov 1, PhD, Michael J Owen 1, PhD, Michael C O Donovan 1, PhD 1 - MRC Centre for Neuropsychiatric Genetics & Genomics, Cardiff Univewrsity, UK * - corresponding author e-mail: EscottPriceV@cardiff.ac.uk Phone: +442920688429 Address: Hadyn Ellis Building, Cardiff University, Maindy Road, Cardiff, UK, CF24 4HQ Running title: Liability to schizophrenia and educational attainment Schizophrenia is associated with impairments in several domains of cognitive function performance and reduced educational achievement (1). Genetic liability conferred by common risk alleles can be directly estimated in individuals regardless of their affected status through a process known as polygenic risk scoring (PRS) (2). PRS have repeatedly been demonstrated to provide a useful index of genetic liability to the disorder (2, 3). Previous reports have shown that (higher) trait liability as defined by higher PRS for schizophrenia associated alleles is associated with reduced performance for several measures of cognitive function in unaffected individuals (4, 5). Despite being associated with poor cognitive performance, schizophrenia PRS has somewhat paradoxically been associated with higher educational attainment (4) and with greater number of years in education (6). Seeking to 1
explore this further, we implemented PRS analysis in the full UK Biobank (an extended version of the sample used in (4)) to determine the relationship between liability to schizophrenia and academic educational attainment. We restricted the sample to those who self-reported as being of white UK or Irish ancestry, as the schizophrenia GWAS which we used to define risk alleles, was of primarily European Ancestry (7). We also excluded a random member of each pair of related individuals with estimated kinship coefficient > 0.4 (N=179) and people with self-reported schizophrenia (N=660), retaining 442,192 individuals. For constructing PRS, we used imputed data based on the Haplotype Reference Consortium (http://www.haplotype-referenceconsortium.org/site). We retained SNPs (N=7,654,308) with minor allele frequencies., imputation quality score 0.4, and Hardy-Weinberg equilibrium p-value -6. The correlated SNPs were pruned using parameters r 2 =0.2, a physical distance threshold of 1Mb, preferentially retaining the SNP most significantly associated with schizophrenia (7). The schizophrenia GWAS association p-value threshold for SNPs inclusion was 0.05 as this currently maximally captures polygenic risk in the greatest number of samples (3). PRS were adjusted for array (Axiom and BiLEVE Affymetrix arrays) and, to allow for ancestry based structure in the data not captured by self-report, for the first 15 principal components, which resulted in multiple clusters of individuals. PRS were further standardised. We included age and sex as covariates in all analyses. Educational achievement in UK Biobank (http://biobank.ctsu.ox.ac.uk/crystal/field.cgi?id=6138) was coded by integer values 1 to 6, higher scores indicating the lower levels of educational attainment (8). To make our analysis more intuitively interpretable, we reversed the order such that higher score corresponds to a higher level of education; 6- College/University degree, 5- A/AS levels or equivalent (advanced 2
academic qualifications taken post-compulsory education age), 4- O levels/gcse levels or equivalent (largely academic qualifications taken at the end of compulsory education), 3- CSEs or equivalent (certification of at the end of compulsory education, less stringent than O levels), 2- NVQ/HND/HNC (vocational qualifications), 1- none of the a o e. The individuals ith othe p ofessional ualifi ations only and those who did not answer were excluded. The primary analysis of educational achievement was ordinal regression, exploratory binary analyses were tested using logistic regression. The CSE was introduced in 1965 so we adjusted our analyses with a binary variable, indicating those born before or after 1950. Higher genetic loading to schizophrenia was weakly associated with educational attainment when analysed using ordinal regression model (B=-0.010, 95%CI=[-0.016, -0.005], p=1.96x10-4 ), the negative B-coefficient indicating higher SZ PRS is correlated lower educational attainment. Adjusting for birth before or after 1950 did not change the results (p=2.83x10-4 ). To compare our findings with (4), we compared those with a college or university level degree versus the rest (as in (4)), but found no significant evidence for association (B=0.006, 95%CI=[- 0.0007, 0.012], p=0.082; here 1, is a college or university level degree, 0 -others). Dichotomizing at the university/college/a/as levels ( 1 ) vs others ( 0 ) was similarly not significant (B=-0.005, 95%CI=[-0.011, 0.002], p=0.149). In contrast, dichotomizing at the largely academic qualifications (university though to GCSEs) versus the rest, resulted in a highly significant association; higher schizophrenia polygenic risk score being associated with an absence of academic qualification (B=-0.026,95%CI=[-0.033, -0.020], p=1.4x10-14 ), see Figure (A). Our finding that schizophrenia liability is negatively correlated with academic educational achievement on an ordinal scale, and on a dichotomous scale at the boundary between academic and non-academic educational achievement is congruent with findings indicating a 3
negative association between SZ PRS and cognitive function including IQ. Previous work indicating a positive association with educational attainment was based on comparing those with a college or university level degree versus the rest (4). In our study that used the full UKBB sample as opposed to the first release of data from 112,151 subjects we found a trend in the same direction but this was not significant. Why the negative correlation is most markedly manifest in qualifications that in the UK, can be considered stepping stones towards university/college education is unclear. Historically, and within the lifetime of most UK Biobank participants, the opportunities for admission to university/college were highly restricted, and it is possible therefore that the highest levels of academic education are a poorer index of innate cognitive ability than success at school where opportunities were somewhat more equally distributed. However, using PRS for IQ (9) as a proxy for innate cognitive ability did not support this explanation, since those attending university had highest IQ PRS (Figure B). We speculate that an intermediate liability to SZ influences another trait (e.g. creativity (6)) without adversely affecting cognitive ability, and when combined with higher trait cognitive ability, the combination favours academic ability. Finally, since the UK Biobank sample is biased towards healthier, wealthier and better educated people, our study tends to miss out on people who are both poorly educated and who have high SZ PRS, but oversample people who are well educated and/or have lower SZ PRS. This would lower the power, but not generate a false positive. Acknowledgements This research was conducted using the UK Biobank resource. UK Biobank was established by the Wellcome Trust, Medical Research Council, Department of Health, Scottish Government and Northwest Regional Development Agency. UK Biobank has also had funding from the Welsh Assembly 4
Government and the British Heart Foundation. The work at Cardiff University was supported by Medical Research Council (MRC) Centre (MR/L010305/1) and Program Grants (G0800509). Disclosures. The authors have nothing to disclose. References 1. Rajji TK, Voineskos AN, Butters MA, Miranda D, Arenovich T, Menon M, et al. (2013): Cognitive performance of individuals with schizophrenia across seven decades: a study using the MATRICS consensus cognitive battery. Am J Geriatr Psychiatry. 21:108-118. 2. International Schizophrenia C, Purcell SM, Wray NR, Stone JL, Visscher PM, O'Donovan MC, et al. (2009): Common polygenic variation contributes to risk of schizophrenia and bipolar disorder. Nature. 460:748-752. 3. Schizophrenia Working Group of the Psychiatric Genomics C (2014): Biological insights from 108 schizophrenia-associated genetic loci. Nature. 511:421-427. 4. Hagenaars SP, Harris SE, Davies G, Hill WD, Liewald DC, Ritchie SJ, et al. (2016): Shared genetic aetiology between cognitive functions and physical and mental health in UK Biobank (N=112 151) and 24 GWAS consortia. Molecular psychiatry. 21:1624-1632. 5. Hubbard L, Tansey KE, Rai D, Jones P, Ripke S, Chambert KD, et al. (2016): Evidence of Common Genetic Overlap Between Schizophrenia and Cognition. Schizophr Bull. 42:832-842. 6. Power RA, Steinberg S, Bjornsdottir G, Rietveld CA, Abdellaoui A, Nivard MM, et al. (2015): Polygenic risk scores for schizophrenia and bipolar disorder predict creativity. Nat Neurosci. 18:953-955. 7. Pardinas AF, Holmans P, Pocklington AJ, Escott-Price V, Ripke S, Carrera N, et al. (2018): Common schizophrenia alleles are enriched in mutation-intolerant genes and in regions under strong background selection. Nature genetics. 50:381-389. 8. Kendall KM, Rees E, Escott-Price V, Einon M, Thomas R, Hewitt J, et al. (2016): Cognitive Performance Among Carriers of Pathogenic Copy Number Variants: Analysis of 152,000 UK Biobank Subjects. Biological psychiatry. 9. Sniekers S, Stringer S, Watanabe K, Jansen PR, Coleman JRI, Krapohl E, et al. (2017): Erratum: Genome-wide association meta-analysis of 78,308 individuals identifies new loci and genes influencing human intelligence. Nature genetics. 49:1558. 5
Figure. Means of standardised PRS per educational attainment category. B-coefficients and p-values represent logistic regression analyses results comparing two groups (under solid horizontal lines). All p-values for IQ PRS comparisons are p<10-300. 6