Blood-Derived DNA Methylation Signatures of Crohn's Disease and Severity of Intestinal Inflammation

Abstract

Background & aims: Crohn's disease is a relapsing and remitting inflammatory disorder with a variable clinical course. Although most patients present with an inflammatory phenotype (B1), approximately 20% of patients rapidly progress to complicated disease, which includes stricturing (B2), within 5 years. We analyzed DNA methylation patterns in blood samples of pediatric patients with Crohn's disease at diagnosis and later time points to identify changes that associate with and might contribute to disease development and progression.

Methods: We obtained blood samples from 164 pediatric patients (1-17 years old) with Crohn's disease (B1 or B2) who participated in a North American study and were followed for 5 years. Participants without intestinal inflammation or symptoms served as controls (n = 74). DNA methylation patterns were analyzed in samples collected at time of diagnosis and 1-3 years later at approximately 850,000 sites. We used genetic association and the concept of Mendelian randomization to identify changes in DNA methylation patterns that might contribute to the development of or result from Crohn's disease.

Results: We identified 1189 5'-cytosine-phosphate-guanosine-3' (CpG) sites that were differentially methylated between patients with Crohn's disease (at diagnosis) and controls. Methylation changes at these sites correlated with plasma levels of C-reactive protein. A comparison of methylation profiles of DNA collected at diagnosis of Crohn's disease vs during the follow-up period showed that, during treatment, alterations identified in methylation profiles at the time of diagnosis of Crohn's disease more closely resembled patterns observed in controls, irrespective of disease progression to B2. We identified methylation changes at 3 CpG sites that might contribute to the development of Crohn's disease. Most CpG methylation changes associated with Crohn's disease disappeared with treatment of inflammation and might be a result of Crohn's disease.

Conclusions: Methylation patterns observed in blood samples from patients with Crohn's disease accompany acute inflammation; with treatment, these change to resemble methylation patterns observed in patients without intestinal inflammation. These findings indicate that Crohn's disease-associated patterns of DNA methylation observed in blood samples are a result of the inflammatory features of the disease and are less likely to contribute to disease development or progression.

Keywords: Children; Epigenetic Alteration; Inflammatory Bowel Disease; Risk Stratification and Identification of Immunogenetic and Microbial Markers of Rapid Disease Progression in Children with Crohn's Disease (RISK) Study.

Figure 1:

Crohn’s disease at diagnosis is associated with methylation changes at 1189 CpG sites in blood. All the ~850 K CpG sites represented by dots are ordered by genomic position per chromosome (x axis). P values (-log₁₀) of site-specific association with Crohn’s disease is shown on y axis. Dots above the blue line represent CpGs reaching epigenome-wide significance (FDR < 0.05). Dots above the red line represent CpGs reaching epigenome-wide significance after Bonferroni correction (n = 114 CpGs).

Figure 2:

Methylation signatures of Crohn’s disease reflect inflammatory status of the patient. (a) For the 1189 Crohn’s disease associated CpGs, estimated effects (n = 164 cases and 74 controls) on Crohn’s disease at diagnosis (x axis) are strongly correlated with the estimated effects (n = 272: 45 non-IBD, 132 at diagnosis and 95 follow-up samples) on plasma CRP levels within the same subjects (y axis). Maroon dots represent Crohn’s disease CpGs that showed significance with plasma CRP (P < 0.05). (b,c) At the 206 (of 218) CRP-associated CpGs in the latest meta-analysis (n = 8863) by the Ligthart et al. (y axis), (b) 199 had effects on Crohn’s disease at diagnosis, and (c) 196 had effects on CRP, in the same direction in our data. Maroon dots are CpGs from that showed significance with (b) CD and (c) CRP in our data (P < 0.05).

Figure 3:

Disrupted methylation patterns during the diagnosis of Crohn’s disease revert back during the course of the disease. (a) Estimated methylation differences between diagnosis (n = 164) and follow-up (n = 164; y axis) were of similar magnitude to baseline differences between cases (n = 164) and controls (n = 164; x axis). All ~850K CpG sites are shown; the correlation coefficient and P value is for the 1189 Crohn’s disease CpGs that are colored maroon. (b) Same comparison for patients (n = 55) who received an initial diagnosis of B1 at the time of diagnosis and progressed to B2 during the course of the follow-up period. (c) Same comparison for patients (n = 95) who started and remained as B1 during diagnosis and follow-up.

Figure 4:

Boxplots demonstrating the impact of the class of medications on methylation beta values at the top-five Crohn’s disease associated CpGs during follow-up. Methylation levels for the same CpGs in non-IBD controls and Crohn’s disease patients at diagnosis are also shown.

Figure 5:

Evaluation of directionality among Crohn’s disease associated CpG sites. Schematic diagram and results of genetic association and the concept of Mendelian randomization framework implemented to clarify the causal versus consequential role of Crohn’s disease associated methylation changes in blood. (a) Overall strength of causality of the 174 CpGs tested for methylation cause of Crohn’s disease is inferred from the correlation between the observed (y axis) and predicted effects (x axis) of SNP on Crohn’s disease. To infer causality of individual CpG sites, the association of the sentinel mQTL with Crohn’s disease should be significant (FDR < 0.05). 95% CI error bars are shown for the 3 CpG sites with an associated mQTL that also associated with Crohn’s disease. 82 of the 174 CpGs that demonstrated directional consistency are shown in maroon; CpGs that are directionally inconsistent with the observed versus predicted effects are shown in grey. (b) Observed effect of Crohn’s disease genetic risk score on methylation (y axis) is highly correlated with its predicted effect (x axis) through Crohn’s disease, suggesting a strong consequential signal at the 194 CpG sites investigated. 95% CI error bars are shown for 8 CpGs demonstrating statistically significant consequential association (FDR < 0.05). 142 of the 194 CpGs with directionally consistent effects are shown in maroon; CpGs that are directionally inconsistent are shown in grey.