Search

ObjectivePsychosis - a complex and heterogeneous neuropsychiatric condition characterized by hallucinations and delusions - is a common feature of schizophrenia. There is evidence for altered DNA methylation (DNAm) associated with schizophrenia in both brain and peripheral tissues. We aimed to undertake a systematic analysis of variable DNAm associated with psychosis, schizophrenia, and treatment-resistant schizophrenia, also exploring measures of biological ageing, smoking, and blood cell composition derived from DNAm data to identify molecular biomarkers of disease.MethodsWe quantified DNAm across the genome in blood samples from 4,483 participants from seven case-control cohorts including patients with schizophrenia or first-episode psychosis. Measures of biological age, cellular composition and smoking status were derived from DNAm data using established algorithms. DNAm and derived measures were analyzed within each cohort and the results combined by meta-analysis.ResultsPsychosis cases were characterized by significant differences in measures of blood cell proportions and elevated smoking exposure derived from the DNAm data, with the largest differences seen in treatment-resistant schizophrenia patients. DNAm at 95 CpG sites was significantly different between psychosis cases and controls, with 1,048 differentially methylated positions (DMPs) identified between schizophrenia cases and controls. Schizophrenia-associated DMPs colocalize to regions identified in genetic association studies, with genes annotated to these sites enriched for pathways relevant to disease. Finally, a number of the schizophrenia associated differences were only present in the treatment-resistant schizophrenia subgroup.ConclusionsWe show that DNAm data can be leveraged to derive measures of blood cell counts and smoking that are strongly associated with psychosis. Our DNAm meta-analysis identified multiple DMPs associated with both psychosis and a more refined diagnosis of schizophrenia, with evidence for differential methylation associated with treatment-resistant schizophrenia that potentially reflects exposure to clozapine.

Table S1. Estimating the multiple testing correction significance threshold for sub-samples of EPIC array DNA methylation sites. Table S2. Summary of results from tests of assumptions of linear regression separated by mean DNA methylation level. The number and percentage of DNA methylation sites significant for each test (P

Figure S1. Distribution of minimum p-values from 1000 null EWAS simulations. The red line at 9.42 × 10–8 is the 5th percentile point and represents the 5% FWER. Figure S2. The four assumptions of linear regression (described in the left boxes), were tested using R package gvlma with four statistical tests (described in the right boxes) with arrows matching the assumptions to the relevant test. Gvlma calculates a p-value for each test, where the null hypothesis is that the assumption(s) hold true, and the alternative hypothesis is that they do not (i.e. a significant p-value indicates that the assumption(s) are violated). Figure S3. Quantile-Quantile plots of the 5 tests of the assumptions of linear regression. Plotted are the observed (x-axis) against expected (y-axis) –log10(p-value) from the (a) global (b) skewness (c) kurtosis (d) link function and (e) heteroskedasticity tests performed in the R gvlma package for all DNA methylation sites. Under the null distribution, of no significant associations, all points would be expected to lie on the red line at y = x. The observed data show a dramatic inflation of p-values smaller than expected by chance in all 5 plots indicating that many DNA methylation sites fail these statistical tests for the assumptions of linear regression. Figure S4. Scatterplots of –log10(p-value) against mean DNA methylation level from the (a) global (b) skewness (c) kurtosis (d) link function and (e) heteroskedasticity tests performed in the R gvlma package for all DNA methylation sites. Each point represents a single site, and the color of the point represents the density of points plotted (low density in grey to high density in yellow). Figure S5. Scatterplots of DNA methylation standard deviation against –log10(p-values) from the (a) global (b) skewness (c) kurtosis (d) link function and (e) heteroskedasticity tests performed in the R gvlma package. Each point represents a single DNA methylation site, and the color of the points represents the density of points plotted (low density in grey to high density in yellow). Figure S6. Boxplots of –log10(p-value) for each of the 5 tests: (a) global (b) skewness (c) kurtosis (d) link function and (e) heteroscedasticity separated by site variability status. DNA methylation sites were defined as variable if the range of their middle 80% of values, calculated as the 90th percentile (P90) minus the 10th percentile (P10) was greater than 5%. Each boxplot is colored by their mean –log10(p-value), from light yellow (lowest –log10(p-value)) to red (highest –log10(p-value). Sites with a p-value of 0 (i.e. p