Systematic Review and Meta-analysis of Peripheral Blood DNA Methylation Studies in Inflammatory Bowel Disease

Abstract

Background and aims: Over the past decade, the DNA methylome has been increasingly studied in peripheral blood of inflammatory bowel disease [IBD] patients. However, a comprehensive summary and meta-analysis of peripheral blood leukocyte [PBL] DNA methylation studies has thus far not been conducted. Here, we systematically reviewed all available literature up to February 2022 and summarized the observations by means of meta-analysis.

Methods: We conducted a systematic search and critical appraisal of IBD-associated DNA methylation studies in PBL using the biomarker-based cross-sectional studies [BIOCROSS] tool. Subsequently, we performed meta-analyses on the summary statistics obtained from epigenome-wide association studies [EWAS] that included patients with Crohn's disease [CD], ulcerative colitis [UC] and/or healthy controls [HC].

Results: Altogether, we included 15 studies for systematic review. Critical appraisal revealed large methodological and outcome heterogeneity between studies. Summary statistics were obtained from four studies based on a cumulative 552 samples [177 CD, 132 UC and 243 HC]. Consistent differential methylation was identified for 256 differentially methylated probes [DMPs; Bonferroni-adjusted p ≤ 0.05] when comparing CD with HC and 103 when comparing UC with HC. Comparing IBD [CD + UC] with HC resulted in 224 DMPs. Importantly, several of the previously identified DMPs, such as VMP1/TMEM49/MIR21 and RPS6KA2, were consistently differentially methylated across all studies.

Conclusion: Methodological homogenization of IBD epigenetic studies is needed to allow for easier aggregation and independent validation. Nonetheless, we were able to confirm previous observations. Our results can serve as the basis for future IBD epigenetic biomarker research in PBL.

Keywords: DNA methylation; biomarkers; epigenetics; inflammatory bowel diseases; systematic review and meta-analysis.

Figure 1.

Study selection. Studies obtained through database searches were initially screened based on abstract exclusion criteria, followed by full-text exclusion criteria and final meta-analysis exclusion criteria.

Figure 2.

Meta-analysis of the comparison Crohn’s disease [CD] with healthy controls [HC] using data from Harris et al. 2012, Adams et al. 2014, Li Yim et al. 2016 and Ventham et al. 2016. [A] A Manhattan plot representing the statistical significance depicted as −log₁₀[p-value] on the y-axis relative to the location across the genome [hg19] on the x-axis. Red and green dots represent the five most significantly hyper- and hypomethylated differentially methylated probes [DMPs], respectively. [B] Forest plots representing the standardized effect size [SES] of interest per study of the five most significantly hyper- and hypomethylated DMPs annotated with the associated gene and p-value from the meta-analysis. [C] Barchart representing the top 25 most significant differentially methylated genes [DMGs]. The length of the bar is proportional to the −log₁₀[p-value] obtained from the DMG analysis. [D] Genomic visualization and forest plots of the estimated difference in methylation for the DMPs cg16936953, cg12054453 and cg18942579, all of which have been associated with VMP1/TMEM49/MIR21. [E] Forest plots of DMPs cg17501210 and cg18608055, associated with RPS6KA2 and SBNO2, respectively, which had been reported on in multiple earlier studies.

Figure 3.

Meta-analysis of the comparison ulcerative colitis [UC] with healthy controls [HC] using data from Harris et al. 2012 and Ventham et al. 2016. [A] A Manhattan plot representing the statistical significance depicted as −log₁₀[p-value] on the y-axis relative to the location across the genome [hg19] on the x-axis. Red and green dots represent the five most significantly hyper- and hypomethylated differentially methylated probes [DMPs], respectively. [B] Forest plots representing the standardized effect size [SES] of interest per study of the five most hyper- and hypomethylated DMPs annotated with the associated gene and p-value from the meta-analysis. [C] Barchart representing the top 25 most significant differentially methylated genes [DMGs]. The length of the bar is proportional to the −log₁₀[p-value] obtained from the DMG analysis.

Figure 4.

Meta-analysis of the comparisons Crohn’s disease [CD] with ulcerative colitis [UC] and inflammatory bowel disease [IBD] with healthy controls [HC] using data from Harris et al. 2012 and Ventham et al. 2016. [A] A scatter plot of the included CpGs representing aggregated effect size of CD with HC and UC with HC on the x- and y-axis, respectively. Colour intensity is proportional to the distance of each CpG to the origin [0,0]. Annotated are the CpGs that are concordantly hyper- and hypomethylated in both CD and UC, and hence IBD, relative to HC. [B] Forest plots representing the standardized effect size [SES] of interest per study of the five most hyper- and hypomethylated IBD-associated DMPs annotated with the associated gene and p-value from the meta-analysis. Genomic visualization and forest plots of the estimated difference in methylation for the CD/UC discordant genes [C] IFITM1 and [D] ZNF875, alongside for the most significant differentially methylated probes.